DWM Lab Manual

(15A05610) DATA WAREHOUSING & MINING LABORATORY
Data Warehousing
Experiments:
1. Build Data Warehouse and Explore WEKA
A. Build a Data Warehouse/Data Mart (using open source tools like Pentaho Data
Integration tool, Pentaho Business Analytics; or other data warehouse tools like Microsoft-
SSIS, Informatica, Business Objects, etc.).
(i). Identify source tables and populate sample data
(ii).Design multi-dimensional data models namely Star, Snowflake and Fact constellation
schemas for any one enterprise (ex. Banking, Insurance, Finance, Healthcare, Manufacturing,
Automobile, etc.).
(iii). Write ETL scripts and implement using data warehouse tools
(iv). Perform various OLAP operations such slice, dice, roll up, drill down and pivot
(v). Explore visualization features of the tool for analysis like identifying trends etc.
B. Explore WEKA Data Mining/Machine Learning Toolkit
(i). Downloading and/or installation of WEKA data mining toolkit,
(ii). Understand the features of WEKA toolkit such as Explorer, Knowledge Flow interface,
Experimenter, command-line interface.
(iii). Navigate the options available in the WEKA (ex. Select attributes panel, Preprocess panel,
classify panel, Cluster panel, Associate panel and Visualize panel)
(iv). Study the arff file format
(v). Explore the available data sets in WEKA.
(vi). Load a data set (ex. Weather dataset, Iris dataset, etc.)
(vii). Load each dataset and observe the following:
i. List the attribute names and they types
ii. Number of records in each dataset
iii. Identify the class attribute (if any)
iv. Plot Histogram
v. Determine the number of records for each class.
vi. Visualize the data in various dimensions
2. Perform data preprocessing tasks and Demonstrate performing association rule
mining on data sets
A. Explore various options available in Weka for preprocessing data and apply (like
Discretization Filters, Resample filter, etc.) on each dataset
B. Load each dataset into Weka and run Aprori algorithm with different support and confidence
values. Study the rules generated.
C. Apply different discretization filters on numerical attributes and run the Apriori association
rule algorithm. Study the rules generated. Derive interesting insights and observe the effect of
discretization in the rule generation process.
3. Demonstrate performing classification on data sets
A. Load each dataset into Weka and run Id3, J48 classification algorithm. Study the classifier
output. Compute entropy values, Kappa statistic.
B. Extract if-then rules from the decision tree generated by the classifier, Observe the confusion
matrix and derive Accuracy, F-measure, TPrate, FPrate, Precision and Recall values. Apply
cross-validation strategy with various fold levels and compare the accuracy results.
C. Load each dataset into Weka and perform Naïve-bayes classification and k- Nearest
Neighbour classification. Interpret the results obtained.
D. Plot RoC Curves
E. Compare classification results of ID3, J48, Naïve-Bayes and k-NN classifiers for each dataset,
and deduce which classifier is performing best and poor for each dataset and justify.
4. Demonstrate performing clustering on data sets
A. Load each dataset into Weka and run simple k-means clustering algorithm with different
values of k (number of desired clusters). Study the clusters formed. Observe the sum of squared
errors and centroids, and derive insights.
B. Explore other clustering techniques available in Weka.
C. Explore visualization features of Weka to visualize the clusters. Derive interesting insights
and explain.
5. Demonstrate performing Regression on data sets
A. Load each dataset into Weka and build Linear Regression model. Study the clusters formed.
Use Training set option. Interpret the regression model and derive patterns and conclusions from
the regression results.
B. Use options cross-validation and percentage split and repeat running the Linear Regression
Model. Observe the results and derive meaningful results.
C. Explore Simple linear regression technique that only looks at one variable
Resource Sites:
1. http://www.pentaho.com/
2. http://www.cs.waikato.ac.nz/ml/weka/
Data Mining
Task 1: Credit Risk Assessment
Description:
The business of banks is making loans. Assessing the credit worthiness of an applicant is of
crucial importance. You have to develop a system to help a loan officer decide whether the credit
of a customer is good, or bad. A bank's business rules regarding loans must consider two
opposing factors. On the one hand, a bank wants to make as many loans as possible. Interest on
these loans is the banks profit source. On the other hand, a bank cannot afford to make too many
bad loans. Too many bad loans could lead to the collapse of the bank. The bank's loan policy
must involve a compromise: not too strict, and not too lenient.
To do the assignment, you first and foremost need is some knowledge about the world of credit.
You can acquire such knowledge in a number of ways
1. Knowledge Engineering. Find a loan officer who is willing to talk. Interview her and try to
represent her knowledge in the form of production rules.
2. Books. Find some training manuals for loan officers or perhaps a suitable textbookon finance.
Translate this knowledge from text form to production rule form.
3. Common sense. Imagine yourself as a loan officer and make up reasonable rules which can be
used to judge the credit worthiness of a loan applicant.
4. Case histories. Find records of actual cases where competent loan officerscorrectly judged
when, and when not to, approve a loan application.
The German Credit Data:
Actual historical credit data is not always easy to come by because of confidentialityrules. Here
is one such dataset, consisting of 1000 actual cases collected in Germany. credit dataset
(original) Excel spreadsheet version of the German credit data. In spite of the fact that the data is
German, you should probably make use of it for this assignment. (Unless you really can consult a
real loan officer!)
A few notes on the German dataset
•DM stands for Deutsche Mark, the unit of currency, worth about 90 cents Canadian (but looks
and acts like a quarter).
•Owns_telephone. German phone rates are much higher. So fewer people own telephones.
•Foreign_worker. There are millions of these in Germany (many from Turrkey). It is very hard to
get German citizenship if you were not born of German parents.
•There are 20 attributes used in judging a loan applicant. The goal is to classify the applicant into
one of two categories, good or bad.
Subtasks: (Turn in your answers to the following tasks)
1. List all the categorical (or nominal) attributes and the real-valued attributesseparately.
2. What attributes do you think might be crucial in making the credit assessment? Come up with
some simple rules in plain English using your selected attributes.
3. One type of model that you can create is a Decision Tree - train a Decision Tree using the
complete dataset as the training data. Report the model obtained after training.
4. Suppose you use your above model trained on the complete dataset, and classify credit
good/bad for each of the examples in the dataset. What % of examples can you classify
correctly? (This is also called testing on the training set) Why do you think you cannot get 100 %
training accuracy?
5. Is testing on the training set as you did above a good idea? Why or Why not?
6. One approach for solving the problem encountered in the previous question is using cross-
validation? Describe what is cross-validation briefly. Train a Decision Tree again using cross-
validation and report your results. Does your accuracy increase/decrease? Why?
7. Check to see if the data shows a bias against "foreign workers" (attribute 20), or"personal-
status" (attribute 9). One way to do this (perhaps rather simple minded) is to remove these
attributes from the dataset and see if the decision tree created in those cases is significantly
different from the full dataset case which you have already done. To remove an attribute, you can
use the preprocess tab in Weka's GUI Explorer. Did removing these attributes have any
significant effect?
8. Another question might be, do you really need to input so many attributes to getgood results?
Maybe only a few would do. For example, you could try just having attributes 2, 3, 5, 7, 10, 17
(and 21, the class attribute (naturally)). Try out some combinations. (You had removed two
attributes in problem 7. Remember to reload the arff data file to get all the attributes initially
before you start selecting the ones you want.)
9. Sometimes, the cost of rejecting an applicant who actually has a good credit (case1) might be
higher than accepting an applicant who has bad credit (case 2). Instead of counting the
misclassifications equally in both cases, give a higher cost to the first case (say cost 5) and lower
cost to the second case. You can do this by using a cost matrix in Weka. Train your Decision
Tree again and report the Decision Tree and crossvalidation results. Are they significantly
different from results obtained in problem 6 (using equal cost)?
10. Do you think it is a good idea to prefer simple decision trees instead of having longcomplex
decision trees? How does the complexity of a Decision Tree relate to the bias of the model?
11. You can make your Decision Trees simpler by pruning the nodes. One approach is to use
Reduced Error Pruning. Try reduced error pruning for training your Decision Trees using cross-
validation (you can do this in Weka) and report the Decision Tree you obtain? Also, report your
accuracy using the pruned model. Does your accuracy increase?
12.(Extra Credit): How can you convert a Decision Trees into "if-then-else rules". Make up your
own small Decision Tree consisting of 2-3 levels and convert it into a set of rules. There also
exist different classifiers that output the model in the form of rules - one such classifier in Weka
is rules. PART, train this model and report the set of rules obtained. Sometimes just one attribute
can be good enough in making the decision, yes, just one ! Can you predict what attribute that
might be in this dataset? OneR classifier uses a single attribute to make decisions (it chooses the
attribute based on minimum error). Report the rule obtained by training a one R classifier. Rank
the performance of j48, PART and oneR.
Task 2: Hospital Management System
Data Warehouse consists Dimension Table and Fact Table.
REMEMBER The following
Dimension
The dimension object (Dimension):
_ Name
_ Attributes (Levels) , with one primary key
_ Hierarchies
One time dimension is must.
About Levels and Hierarchies
Dimension objects (dimension) consist of a set of levels and a set of hierarchies defined over
those levels. The levels represent levels of aggregation. Hierarchies describe parent-child
relationships among a set of levels.
For example, a typical calendar dimension could contain five levels. Two hierarchies
can be defined on these levels:
H1: YearL > QuarterL > MonthL > WeekL > DayL
H2: YearL > WeekL > DayL
The hierarchies are described from parent to child, so that Year is the parent of Quarter, Quarter
the parent of Month, and so forth.
About Unique Key Constraints
When you create a definition for a hierarchy, Warehouse Builder creates an identifier key for
each level of the hierarchy and a unique key constraint on the lowest level (Base Level)
Design a Hospital Management system data warehouse (TARGET) consisting of Dimensions
Patient, Medicine, Supplier, Time. Where measures are ‗ NO UNITS ‘,
UNIT PRICE.
Assume the Relational database (SOURCE) table schemas as follows
TIME (day, month, year),
PATIENT (patient_name, Age, Address, etc.,)
MEDICINE ( Medicine_Brand_name, Drug_name, Supplier, no_units, Uinit_Price, etc.,)
SUPPLIER :( Supplier_name, Medicine_Brand_name, Address, etc., )
If each Dimension has 6 levels, decide the levels and hierarchies, Assume the level names
suitably.
Design the Hospital Management system data warehouse using all schemas. Give the example 4-
D cube with assumption names.
Data Warehousing
Experiments:
1. Build Data Warehouse and Explore WEKA
A. Build a Data Warehouse/Data Mart (using open source tools like Pentaho Data
Integration tool, Pentaho Business Analytics; or other data warehouse tools like Microsoft-
SSIS, Informatica, Business Objects, etc.).
(i). Identify source tables and populate sample data
In this task, we are going to use MySQL administrator, SQLyog Enterprise tools for
building & identifying tables in database & also for populating (filling) the sample data in those
tables of a database.A data warehouse is constructed by integrating data from multiple
heterogeneous sources. It supports analytical reporting, structured and/or ad hoc queries and
decision making. We are building a data warehouse by integrating all the tables in database &
analyzing those data. In the below figure we represented MySQL Administrator connection
establishment.
After successful login, it will open new window as shown below.

There are different options available in MySQL administrator. Another tool SQLyog
Enterprise, we are using for building & identifying tables in a database after successful
connection establishment through MySQL Administrator. Below we can see the window of
SQLyog Enterprise.
On left-side navigation, we can see different databases & it’s related tables. Now we are
going to build tables & populate table’s data in database through SQL queries. These tables in
database can be used further for building data warehouse.
In the above two windows, we created a database named “sample”& in that database we
created two tables named as “user_details”& “hockey”through SQL queries.
Now, we are going to populate (filling) sample data through SQL queries in those two
created tables as represented in below windows.
Through MySQL administrator & SQLyog, we can import databases from other sources
(.XLS, .CSV, .sql) & also we can export our databases as backup for further processing. We can
connect MySQL to other applications for data analysis & reporting.
(ii). Design multi-dimensional data models namely Star, snowflake and Fact constellation
schemas for any one enterprise (ex. Banking, Insurance, Finance, Healthcare,
Manufacturing, Automobile, etc.).
Multi-Dimensional model was developed for implementing data warehouses & it provides both a
mechanism to store data and a way for business analysis. The primary components of
dimensional model are dimensions & facts. There are different of types of multi-dimensional
data models. They are:
1. Star Schema Model
2. Snow Flake Schema Model
3. Fact Constellation Model.
Now, we are going to design these multi-dimensional models for the Marketing
enterprise.
First, we need to built the tables in a database through SQLyog as shown below.
In the above window, left side navigation bar consists of a database named as “sales_dw”
in which there are six different tables (dimcustdetails, dimcustomer, dimproduct, dimsalesperson,
dimstores, factproductsales) has been created.
After creating tables in database, here we are going to use a tool called as “Microsoft
Visual Studio 2012 for Business Intelligence” for building multi-dimensional models.
In the above window, we are seeing Microsoft Visual Studio before creating a project In
which right side navigation bar contains different options like Data Sources, Data Source Views,
Cubes, Dimensions etc.
Through Data Sources, we can connect to our MySQL database named as “sales_dw”.
Then, automatically all the tables in that database will be retrieved to this tool for creating multi-
dimensional models.
By data source views & cubes, we can see our retrieved tables in multi-dimensional
models. We need to add dimensions also through dimensions option. In general, Multi-
dimensional models consists of dimension tables & fact tables.
Star Schema Model:
A Star schema model is a join between a fact table and a no. of dimension tables. Each
dimensional table are joined to the fact table using primary key to foreign key join but
dimensional tables are not joined to each other. It is the simplest style of dataware house schema.
Star schema is a entity relationship diagram of this schema resembles a star with point
radiating from central table as we seen in the below implemented window in visual studio.
Snow Flake Schema:
It is slightly different from star schema in which dimensional tables from a star schema
are organized into a hierarchy by normalizing them.
Snow flake schema is represented by centralized fact table which are connected to
multiple dimension tables. Snow flake effects only dimension tables not fact tables. we
developed a snowflake schema for sales_dw database by visual studio tool as shown below.
Fact Constellation Schema:
Fact Constellation is a set of fact tables that share some dimension tables. In this schema
there are two or more fact tables. We developed fact constellation in visual studio as shown
below. Fact tables are labelled in yellow color.
(iii). Write ETL scripts and implement using data warehouse tools
ETL (Extract-Transform-Load):
ETL comes from Data Warehousing and stands for Extract-Transform-Load. ETL covers a
process of how the data are loaded from the source system to the data warehouse. Currently, the
ETL encompasses a cleaning step as a separate step. The sequence is then Extract-Clean-
Transform-Load. Let us briefly describe each step of the ETL process.
Process
Extract:
The Extract step covers the data extraction from the source system and makes it accessible for
further processing. The main objective of the extract step is to retrieve all the required data from
the source system with as little resources as possible. The extract step should be designed in a
way that it does not negatively affect the source system in terms or performance, response time
or any kind of locking.
There are several ways to perform the extract:
● Update notification - if the source system is able to provide a notification that a record
has been changed and describe the change, this is the easiest way to get the data.
● Incremental extract - some systems may not be able to provide notification that an update
has occurred, but they are able to identify which records have been modified and provide
an extract of such records. During further ETL steps, the system needs to identify
changes and propagate it down. Note, that by using daily extract, we may not be able to
handle deleted records properly.
● Full extract - some systems are not able to identify which data has been changed at all, so
a full extract is the only way one can get the data out of the system. The full extract
requires keeping a copy of the last extract in the same format in order to be able to
identify changes. Full extract handles deletions as well.
When using Incremental or Full extracts, the extract frequency is extremely important.
Particularly for full extracts; the data volumes can be in tens of gigabytes.
Clean:
The cleaning step is one of the most important as it ensures the quality of the data in the data
warehouse. Cleaning should perform basic data unification rules, such as:
● Making identifiers unique (sex categories Male/Female/Unknown, M/F/null,
Man/Woman/Not Available are translated to standard Male/Female/Unknown)
● Convert null values into standardized Not Available/Not Provided value
● Convert phone numbers, ZIP codes to a standardized form
● Validate address fields, convert them into proper naming, e.g. Street/St/St./Str./Str
● Validate address fields against each other (State/Country, City/State, City/ZIP code,
City/Street).
Transform:
The transform step applies a set of rules to transform the data from the source to the target. This
includes converting any measured data to the same dimension (i.e. conformed dimension) using
the same units so that they can later be joined. The transformation step also requires joining data
from several sources, generating aggregates, generating surrogate keys, sorting, deriving new
calculated values, and applying advanced validation rules.
Load:
During the load step, it is necessary to ensure that the load is performed correctly and with as
little resources as possible. The target of the Load process is often a database. In order to make
the load process efficient, it is helpful to disable any constraints and indexes before the load and
enable them back only after the load completes. The referential integrity needs to be maintained
by ETL tool to ensure consistency.
Managing ETL Process:
The ETL process seems quite straight forward. As with every application, there is a possibility
that the ETL process fails. This can be caused by missing extracts from one of the systems,
missing values in one of the reference tables, or simply a connection or power outage. Therefore,
it is necessary to design the ETL process keeping fail-recovery in mind.
Staging:
It should be possible to restart, at least, some of the phases independently from the others. For
example, if the transformation step fails, it should not be necessary to restart the Extract step. We
can ensure this by implementing proper staging. Staging means that the data is simply dumped to
the location (called the Staging Area) so that it can then be read by the next processing phase.
The staging area is also used during ETL process to store intermediate results of processing. This
is ok for the ETL process which uses for this purpose. However, tThe staging area should is be
accessed by the load ETL process only. It should never be available to anyone else; particularly
not to end users as it is not intended for data presentation to the end-user.may contain incomplete
or in-the-middle-of-the-processing data.
ETL Tool Implementation:

When you are about to use an ETL tool, there is a fundamental decision to be made: will the
company build its own data transformation tool or will it use an existing tool?
Building your own data transformation tool (usually a set of shell scripts) is the preferred
approach for a small number of data sources which reside in storage of the same type. The reason
for that is the effort to implement the necessary transformation is little due to similar data
structure and common system architecture. Also, this approach saves licensing cost and there is
no need to train the staff in a new tool. This approach, however, is dangerous from the TOC
point of view. If the transformations become more sophisticated during the time or there is a
need to integrate other systems, the complexity of such an ETL system grows but the
manageability drops significantly. Similarly, the implementation of your own tool often
resembles re-inventing the wheel.
There are many ready-to-use ETL tools on the market. The main benefit of using off-the-shelf
ETL tools is the fact that they are optimized for the ETL process by providing connectors to
common data sources like databases, flat files, mainframe systems, xml, etc. They provide a
means to implement data transformations easily and consistently across various data sources.
This includes filtering, reformatting, sorting, joining, merging, aggregation and other operations
ready to use. The tools also support transformation scheduling, version control, monitoring and
unified metadata management. Some of the ETL tools are even integrated with BI tools.
Some of the Well Known ETL Tools:
The most well-known commercial tools are Ab Initio, IBM InfoSphere
DataStage, Informatica, Oracle Data Integrator, and SAP Data Integrator.
There are several open source ETL tools are OpenRefine,
Apatar, CloverETL, Pentaho and Talend.
In these above tools, we are going to use OpenRefine 2.8 ETL toolto different sample datasets
forextracting, data cleaning, transforming & loading.
(iv). Perform various OLAP operations such slice, dice, roll up, drill down and pivot.
OLAP Operations:
Since OLAP servers are based on multidimensional view of data, we will discuss OLAP
operations in multidimensional data.
Here is the list of OLAP operations
● Roll-up (Drill-up)
● Drill-down
● Slice and dice
● Pivot (rotate)
Roll-up (Drill-up):
Roll-up performs aggregation on a data cube in any of the following ways
● By climbing up a concept hierarchy for a dimension
● By dimension reduction
● Roll-up is performed by climbing up a concept hierarchy for the dimension location.
● Initially the concept hierarchy was "street < city < province < country".
● On rolling up, the data is aggregated by ascending the location hierarchy from the level of
city to the level of country.
● The data is grouped into cities rather than countries.
● When roll-up is performed, one or more dimensions from the data cube are removed.
Drill-down:
Drill-down is the reverse operation of roll-up. It is performed by either of the following ways
● By stepping down a concept hierarchy for a dimension
● By introducing a new dimension.
● Drill-down is performed by stepping down a concept hierarchy for the dimension time.
● Initially the concept hierarchy was "day < month < quarter < year."
● On drilling down, the time dimension is descended from the level of quarter to the level
of month.
● When drill-down is performed, one or more dimensions from the data cube are added.
● It navigates the data from less detailed data to highly detailed data.
Slice:
The slice operation selects one particular dimension from a given cube and provides a
new sub-cube.
Dice:
Dice selects two or more dimensions from a given cube and provides a new sub-cube.
Pivot (rotate):
The pivot operation is also known as rotation. It rotates the data axes in view in order to
provide an alternative presentation of data.
Now, we are practically implementing all these OLAP Operations using Microsoft
Excel.
Procedure for OLAP Operations:
1.Open Microsoft Excel, go toData tab in top & click on “Existing Connections”.
2. Existing Connections window will be opened, there “Browse for more”option should be
clicked for importing .cub extension file for performing OLAP Operations. For sample, I took
music.cub file.
3.As shown in above window, select “PivotTable Report” and click “OK”.
4.We got all the music.cub data for analyzing different OLAP Operations.Firstly, we performed
drill-down operation as shown below.
In the above window, we selected year ‘2008’ in ‘Electronic’ Category, then
automatically the Drill-Down option is enabled on top navigation options. We will click on
‘Drill-Down’ option, then the below window will be displayed.
5. Now we are going to perform roll-up (drill-up) operation, in the above window I selected
January month then automatically Drill-up option is enabled on top. We will click on Drill-up
option, then the below window will be displayed.
6. Next OLAP operation Slicing is performed by inserting slicer as shown in top navigation
options.
While inserting slicers for slicing operation, we select 2 Dimensions (for e.g.
CategoryName & Year) only with one Measure (for e.g. Sum of sales).After inserting a slice&
adding a filter (CategoryName: AVANT ROCK & BIG BAND; Year: 2009 & 2010), we will get
table as shown below.
7. Dicing operation is similar to Slicing operation. Here we are selecting 3 dimensions
(CategoryName, Year, RegionCode)& 2 Measures (Sum of Quantity, Sum of Sales) through
‘insert slicer’ option. After that adding a filter for CategoryName, Year & RegionCode as
shown below.
8. Finally, the Pivot (rotate) OLAP operation is performed by swapping rows (Order Date-Year)
& columns (Values-Sum of Quantity & Sum of Sales) through right side bottom navigation bar
as shown below.
After Swapping (rotating), we will get resultant as represented below with a pie-chart for
Category-Classical& Year Wise data.
(v). Explore visualization features of the tool for analysis like identifying trends etc.
There are different visualization features for analyzing the data for trend analysisin data
warehouses. Some of the popular visualizations are:
1. Column Charts
2. Line Charts
3. Pie Chart
4. Bar Graphs
4. Area Graphs
5. X & Y Scatter Graphs
6. Stock Graphs
7. Surface Charts
8. Radar Graphs
9. Treemap
10. Sunburst
11. Histogram
12. Box & Whisker
13. Waterfall
14. Combo Graphs
15. Geo Map
16. Heat Grid
17. Interactive Report
18. Stacked Column
19. Stacked Bar
20. Scatter Area
These type of visualizations can be used for analyzing data for trend analysis. Some of
the tools for data visualization are Microsoft Excel, Tableau, Pentaho Business Analytics Online
etc. Practically different visualization features are tested with different sample datasets.
In the below window, we used 3D-Column Charts of Microsoft Excel for analyzing data
in data warehouse.
Below window, represents the data visualization through Pentaho Business Analytics
tool online (http://www.pentaho.com/hosted-demo) for some sample dataset.
B. Explore WEKA Data Mining/Machine Learning Toolkit
(i). Downloading and/or installation of WEKA data mining toolkit
Procedure:
1. Go to the Weka website, http://www.cs.waikato.ac.nz/ml/weka/, and download the software.
On the left-hand side, click on the link that says download.
2. Select the appropriate link corresponding to the version of the software based on your
operating system and whether or not you already have Java VM running on your machine (if you
don’t know what Java VM is, then you probably don’t).
3. The link will forward you to a site where you can download the software from a mirror site.
Save the self-extracting executable to disk and then double click on it to install Weka. Answer
yes or next to the questions during the installation.
4. Click yes to accept the Java agreement if necessary. After you install the program Weka
should appear on your start menu under Programs (if you are using Windows).
5. Running Weka from the start menu select Programs, then Weka.You will see the Weka GUI
Chooser. Select Explorer. The Weka Explorer will then launch.
(ii). Understand the features of WEKA toolkit such as Explorer, Knowledge Flow
interface, Experimenter, command-line interface.
The Weka GUI Chooser (class weka.gui.GUIChooser) provides a starting pointfor launching
Weka’s main GUI applications and supporting tools. If one prefersa MDI (“multiple document
interface”) appearance, then this is provided by analternative launcher called “Main” (class
weka.gui.Main).
The GUI Chooser consists of four buttons—one for each of the four majorWeka applications—
and four menus.
The buttons can be used to start the following applications:

Explorer- An environment for exploring data with WEKA
a) Click on “explorer” button to bring up the explorer window.
b) Make sure the “preprocess” tab is highlighted.
c) Open a new file by clicking on “Open New file” and choosing a file with “.arff”
extension from the “Data” directory.
d) Attributes appear in the window below.
e) Click on the attributes to see the visualization on the right.
f) Click “visualize all” to see them all
Experimenter- An environment for performing experiments and conducting statistical tests

between learning schemes.
a) Experimenter is for comparing results.
b) Under the “set up” tab click “New”.
c) Click on “Add New” under “Data” frame. Choose a couple of arff format files from
“Data” directory one at a time.
d) Click on “Add New” under “Algorithm” frame. Choose several algorithms, one at a time
by clicking “OK” in the window and “Add New”.
e) Under the “Run” tab click “Start”.
f) Wait for WEKA to finish.
g) Under “Analyses” tab click on “Experiment” to see results.
Knowledge Flow- This environment supports essentially the same functions as the Explorer but
with a drag-and-drop interface. One advantageis that it supports incremental learning.
SimpleCLI - Provides a simple command-line interface that allows directexecution of WEKA
commands for operating systems that do not provide their own command line interface.
(iii). Navigate the options available in the WEKA (ex. Select attributes panel, Preprocess
panel, classify panel, Cluster panel, Associate panel and Visualize panel)
When the Explorer is first started only the first tab is active; the others are greyed out. This is
because it is necessary to open (and potentially pre-process) a data set before starting to explore
the data.
The tabs are as follows:
1. Preprocess. Choose and modify the data being acted on.
2. Classify. Train and test learning schemes that classify or perform regression.
3. Cluster. Learn clusters for the data.
4. Associate. Learn association rules for the data.
5. Select attributes. Select the most relevant attributes in the data.
6. Visualize. View an interactive 2D plot of the data.
Once the tabs are active, clicking on them flicks between different screens, on which the
respective actions can be performed. The bottom area of the window (including the status box,
the log button, and the Weka bird) stays visible regardless of which section you are in.
1. Preprocessing
Loading Data:
The first four buttons at the top of the preprocess section enable you to loaddata into WEKA:
1. Open file.... Brings up a dialog box allowing you to browse for the datafile on the local file
system.
2. Open URL.... Asks for a Uniform Resource Locator address for wherethe data is stored.
3. Open DB.... Reads data from a database. (Note that to make this workyou might have to edit
the file in weka/experiment/DatabaseUtils.props.)
4. Generate.... Enables you to generate artificial data from a variety ofDataGenerators.
Using the Open file... button you can read files in a variety of formats:
WEKA’s ARFF format, CSV format, C4.5 format, or serialized Instances format. ARFF files
typically have a .arff extension, CSV files a .csv extension,C4.5 files a .data and .names
extension, and serialized Instances objects a .bsiextension.
2. Classification:
Selecting a Classifier
At the top of the classify section is the Classifier box. This box has a text fieldthat gives the name of the
currently selected classifier, and its options. Clickingon the text box with the left mouse button brings up
a GenericObjectEditordialog box, just the same as for filters, that you can use to configure the optionsof
the current classifier. With a right click (or Alt+Shift+left click) you canonce again copy the setup string to
the clipboard or display the properties in aGenericObjectEditor dialog box. The Choose button allows
you to choose on4eof the classifiers that are available in WEKA.
Test Options
The result of applying the chosen classifier will be tested according to the optionsthat are set by clicking
in the Test options box. There are four test modes:
1. Use training set: The classifier is evaluated on how well it predicts theclass of the instances it was
trained on.
2. Supplied test set: The classifier is evaluated on how well it predicts theclass of a set of instances
loaded from a file. Clicking the Set... buttonbrings up a dialog allowing you to choose the file to test on.
3. Cross-validation: The classifier is evaluated by cross-validation, usingthe number of folds that are
entered in the Folds text field.
4. Percentage split: The classifier is evaluated on how well it predicts acertain percentage of the data
which is held out for testing. The amountof data held out depends on the value entered in the % field.
3. Clustering:
Cluster Modes:
The Cluster mode box is used to choose what to cluster and how to evaluatethe results. The first three
options are the same as for classification: Use training set, Supplied test set and Percentage split.
4. Associating:
Setting Up
This panel contains schemes for learning association rules, and the learners are chosen and configured in the
same way as the clusterers, filters, and classifiers in the other panels.
5. Selecting Attributes:
Searching and Evaluating

Attribute selection involves searching through all possible combinations of attributes in the data to find
which subset of attributes works best for prediction. To do this, two objects must be set up: an attribute
evaluator and a searchmethod. The evaluator determines what method is used to assign a worth toeach
subset of attributes. The search method determines what style of searchis performed.
6. Visualizing:
WEKA’s visualization section allows you to visualize 2D plots of the current relation.
(iv). Study the arff file format
An ARFF (Attribute-Relation File Format) file is an ASCII text file that describes a list
of instances sharing a set of attributes. ARFF files were developed by the Machine Learning
Project at the Department of Computer Science of The University of Waikato for use with
the Weka machine learning software.
Overview
ARFF files have two distinct sections. The first section is the Header information,
which is followed the Data information.
The Header of the ARFF file contains the name of the relation, a list of the attributes
(the columns in the data), and their types. An example header on the standard IRIS
dataset looks like this:
% 1. Title: Iris Plants Database
%
% 2. Sources:
% (a) Creator: R.A. Fisher
% (b) Donor: Michael Marshall (MARSHALL%[email protected])
% (c) Date: July, 1988
%
@RELATION iris
@ATTRIBUTE sepallength NUMERIC

@ATTRIBUTE sepalwidth NUMERIC
@ATTRIBUTE petallength NUMERIC
@ATTRIBUTE petalwidth NUMERIC
@ATTRIBUTE class {Iris-setosa,Iris-versicolor,Iris-virginica}
The Data of the ARFF file looks like the following:

@DATA
5.1,3.5,1.4,0.2,Iris-setosa
4.9,3.0,1.4,0.2,Iris-setosa
4.7,3.2,1.3,0.2,Iris-setosa
4.6,3.1,1.5,0.2,Iris-setosa
5.0,3.6,1.4,0.2,Iris-setosa
5.4,3.9,1.7,0.4,Iris-setosa
4.6,3.4,1.4,0.3,Iris-setosa
5.0,3.4,1.5,0.2,Iris-setosa
4.4,2.9,1.4,0.2,Iris-setosa
4.9,3.1,1.5,0.1,Iris-setosa
Lines that begin with a % are comments.
The @RELATION, @ATTRIBUTE and @DATA declarations are case insensitive.
(v). Explore the available data sets in WEKA
There are 23 different datasets are available in weka (C:\Program Files\Weka-3-6\) by default for
testing purpose. All the datasets are available in. arff format. Those datasets are listed below.
(vi). Load a data set (ex. Weather dataset, Iris dataset, etc.)
Procedure:
1. Open the weka tool and select the explorer option.
2. New window will be opened which consists of different options (Preprocess, Association etc.)
3. In the preprocess, click the “open file” option.
4. Go to C:\Program Files\Weka-3-6\data for finding different existing. arff datasets.
5. Click on any dataset for loading the data then the data will be displayed as shown below.
(vii). Load each dataset and observe the following:
Here we have taken IRIS.arff dataset as sample for observing all the below things.
i. List the attribute names and they types
There are 5 attributes& its datatype present in the above loaded dataset (IRIS.arff)
sepallength – Numeric
sepalwidth – Numeric
petallength – Numeric
petallength – Numeric
Class – Nominal
ii. Number of records in each dataset
There are total 150 records (Instances) in dataset (IRIS.arff).

iii. Identify the class attribute (if any)
There is one class attribute which consists of 3 labels. They are:

1. Iris-setosa
2. Iris-versicolor
3. Iris-virginica
iv. Plot Histogram
v. Determine the number of records for each class.
There is one class attribute (150 records) which consists of 3 labels. They are shown below
1. Iris-setosa - 50 records
2. Iris-versicolor – 50 records
3. Iris-virginica – 50 records
vi. Visualize the data in various dimensions
2. Perform data preprocessing tasks and Demonstrate performing association rule

mining on data sets
A. Explore various options available in Weka for preprocessing data and apply (like
Discretization Filters, Resample filter, etc.) on each dataset
Procedure:
1. For preprocessing the data after selecting the dataset (IRIS.arff).
2. Select Filter option & apply the resample filter & see the below results.
3. Select another filter option & apply the discretization filter, see the below results
Likewise, we can apply different filters for preprocessing the data & see the results in
different dimensions.
B. Load each dataset into Weka and run Aprori algorithm with different support and
confidence values. Study the rules generated.
Procedure:
1. Load the dataset (Breast-Cancer.arff) into weka tool
2. Go to associate option & in left-hand navigation bar we can see different association
algorithms.
3. In which we can select Aprori algorithm & click on select option.
4. Below we can see the rules generated with different support & confidence values for that
selected dataset.
C. Apply different discretization filters on numerical attributes and run the Apriori
association rule algorithm. Study the rules generated. Derive interesting insights and
observe the effect of discretization in the rule generation process.
Procedure:
1. Load the dataset (Breast-Cancer.arff) into weka tool& select the discretize filter & apply it.
2. Go to associate option & in left-hand navigation bar we can see different association
algorithms.
3. In which we can select Aprori algorithm & click on select option.
4. Below we can see the rules generated with different support & confidence values for that
selected dataset.
3. Demonstrate performing classification on data sets
A. Load each dataset into Weka and run Id3, J48 classification algorithm. Study the
classifier output. Compute entropy values, Kappa statistic.
Procedure for Id3:
1. Load the dataset (Contact-lenses.arff) into weka tool
2. Go to classify option & in left-hand navigation bar we can see differentclassification
algorithms under tree section.
3. In which we selected Id3 algorithm, in more options select the output entropy evaluation
measures& click on start option.
4. Then we will get classifier output, entropy values& Kappa Statistic as represented below.
5. In the above screenshot, we can run classifiers with different test options (Cross-validation,
Use Training Set, Percentage Split, Supplied Test set).
The result of applying the chosen classifier will be tested according to the optionsthat are set by
clicking in the Test options box. There are four test modes:
A. Use training set: The classifier is evaluated on how well it predicts theclass of the instances it
was trained on.
B. Supplied test set: The classifier is evaluated on how well it predicts theclass of a set
ofinstances loaded from a file. Clicking the Set... buttonbrings up a dialog allowing you to
choose the file to test on.
C. Cross-validation: The classifier is evaluated by cross-validation, usingthe number of folds
that are entered in the Folds text field.
D. Percentage split: The classifier is evaluated on how well it predicts acertain percentage of the
data which is held out for testing. The amountof data held out depends on the value entered in
the % field.
Procedure for J48:

1. Load the dataset (Contact-lenses.arff) into weka tool
algorithms under tree section.
3. In which we selected J48 algorithm,in more options select the output entropy evaluation
measures& click on start option.
4. Then we will get classifier output, entropy values & Kappa Statistic as represented below.
5. In the below screenshot, we can run classifiers with different test options (Cross-validation,
Use Training Set, Percentage Split, Supplied Test set).
B. Extract if-then rules from the decision tree generated by the classifier, Observethe
confusion matrix and derive Accuracy, F-measure, TPrate, FPrate, Precision and Recall
values. Apply cross-validation strategy with various fold levels and compare the accuracy
results.
Procedure:
1. Load the dataset (Iris-2D. arff) into weka tool
algorithms under rules section.
3. In which we selected JRip (If-then) algorithm & click on start option with “use training set”
test option enabled.
4. Then we will get detailed accuracy by class consists ofF-measure, TP rate, FP rate, Precision,
Recall values& Confusion Matrix as represented below.
Using Cross-Validation Strategy with 10 folds:
Here, we enabled cross-validation test option with 10 folds & clicked start button as represented
below.
Using Cross-Validation Strategy with 20 folds:
Here, we enabled cross-validation test option with 20 folds & clicked start button as represented
below.
If we see the above results of cross validation with 10 folds & 20 folds. As per our
observation the error rate is lesser with 20 folds got 97.3% correctness when compared to 10
folds got 94.6% correctness.
C. Load each dataset into Weka and perform Naïve-bayes classification and k-Nearest
Neighbour classification. Interpret the results obtained.
Procedure for Naïve-Bayes:

algorithms under bayes section.
3. In which we selected Naïve-Bayes algorithm & click on start option with “use training set”
4. Then we will get detailed accuracy by class consists of F-measure, TP rate, FP rate, Precision,
Procedure for K-Nearest Neighbour (IBK):
algorithms under lazy section.
3. In which we selected K-Nearest Neighbour (IBK) algorithm & click on start option with “use
training set” test option enabled.
D. Plot RoC Curves
Procedure:
3. In which we selected Naïve-Bayes algorithm & click on start option with “use training set”
Recall values& Confusion Matrix.
5. For plotting RoC Curves, we need to right click on “bayes.NaiveBayes” for getting more
options, In which we will select the “Visualize Threshold Curve” & go to any class (Iris-setosa,
Iris-versicolor, Iris-Virgincia) as shown in below snapshot.
6. After selecting an class, RoC (Receiver Operating Characteristic) Curve plot will be displayed
which has X-Axis –False Positive (FP) rate and Y-Axis – True Positive (TP) rate.
E. Compare classification results of ID3, J48, Naïve-Bayes and k-NN classifiers for each
dataset, and deduce which classifier is performing best and poor for each dataset and
justify.
Procedure for ID3:

1. Load the dataset (Contact-Lenses. arff) into weka tool
algorithms under trees section.
3. In which we selected ID3 algorithm & click on start option with “use training set” test option
enabled.
Procedure for J48:

algorithms under trees section.
3. In which we selected J48 algorithm & click on start option with “use training set” test option
enabled.
Procedure for Naïve-Bayes:
2. Go to classify option & in left-hand navigation bar we can see different classification
3. In which we selected NaIve-Bayes algorithm & click on start option with “use training set”
Procedure for K-Nearest Neighbour (IBK):

2. Go to classify option & in left-hand navigation bar we can see differentclassifcation
algorithms under lazy section.
3. In which we selected K-Nearest Neighbour (IBK) algorithm & click on start option with “use
training set” test option enabled.
By observing all these algorithms (ID3, K-NN, J48 & Naïve Bayes) results, we will conclude
that
Hence,
ID3 Algorithm’saccuracy & performance is best.
J48 Algorithm’s accuracy & performance is poor.
4. Demonstrate performing clustering on data sets
A. Load each dataset into Weka and run simple k-means clustering algorithm with
different values of k (number of desired clusters). Study the clusters formed. Observe the
sum of squared errors and centroids, and derive insights.
Procedure:
1. Load the dataset (Iris.arff) into weka tool
2. Go to classify option & in left-hand navigation bar we can see differentclustering algorithms
under lazy section.
3. In which we selected Simple K-Means algorithm & click on start option with “use training
set” test option enabled.
4. Then we will get the sum of squared errors, centroids, No. of iterations & clustered instances
as represented below.
5. If we right click on simple k means, we will get more options in which “Visualize cluster
assignments” should be selected for getting cluster visualization as shown below.
B. Explore other clustering techniques available in Weka.
Clustering:
Selecting a Clusterer:
By now you will be familiar with the process of selecting and configuring objects.Clicking on
the clustering scheme listed in the Clusterer box at the top of thewindow brings up
aGenericObjectEditor dialog with which to choose a newclustering scheme.
Cluster Modes:
The Cluster mode box is used to choose what to cluster and how to evaluatethe results. The first
three options are the same as for classification: Use training set, supplied test set and Percentage
split, now the data is assigned to clusters instead of trying to predict a specific class.The fourth
mode, Classes to clusters evaluation, compares how well thechosen clusters match up with a pre-
assigned class in the data. The drop-downbox below this option selects the class, just as in the
Classify panel.An additional option in the Cluster mode box, the Store clusters forvisualization
tick box, determines whether or not it will be possible to visualizethe clusters once training is
complete. When dealing with datasets that are solarge that memory becomes aproblem it may be
helpful to disable this option.
Ignoring Attributes:
Often, some attributes in the data should be ignored when clustering. TheIgnore attributes button
brings up a small window that allows you to selectwhich attributes are ignored. Clicking on an
attribute in the window highlightsit, holding down the SHIFT key selects a range of consecutive
attributes, andholding down CTRL toggles individual attributes on and off. To cancel
theselection, back out with the Cancel button. To activate it, click the Selectbutton. The next
time clustering is invoked, the selected attributes are ignored.
There are 12 clustering algorithms available in weka tool. They are shown below.
Through visualize cluster assignments, we can clearly see the clusters in graphical visualization.
C. Explore visualization features of Weka to visualize the clusters. Derive interesting

insights and explain.
● If we right click on simple k means, we will get more options in which “Visualize cluster
assignments” should be selected for getting cluster visualization as shown below.
● In that cluster visualization we are having different features to explore by changing the
X-axis, Y-axis, Color, Jitter& Select instance (Rectangle, Polygon & Polyline) for getting
different sets of cluster outputs.
● As shown in above screenshot, all the dataset (Iris.arff) tuples are represented in X-axis &
in similar way it will represented for y-axis also. For each cluster, the color will be
different. In the above figure, there are two clusters which are represented in blue & red
colors.
● In the select instance we can select different shapes for choosing clustered area as shown
in below screenshot, rectangle shape is selected.
● By this visualization feature we can observe different clustering outputs for an dataset by
changing those X-axis, Y-axis, Color & Jitter options.
5. Demonstrate performing Regression on data sets
A. Load each dataset into Weka and build Linear Regression model. Study the clusters
formed. Use Training set option. Interpret the regression model and derive patterns and
conclusions from the regression results.
Procedure:
1. Load the dataset (Cpu.arff) into weka tool
algorithms under functions section.
3. In which we selected Linear Regression algorithm & click on start option with use training set
option.
4. Then we will get regression model & its result as shown below.
5. The patterns are visually mentioned below for regression model through visualize classifier
errors option which is available in right click options.
B. Use options cross-validation and percentage split and repeat running the Linear
Regression Model. Observe the results and derive meaningful results.
Procedure for cross-validation:

3. In which we selected Linear Regression algorithm & click on start option with cross validation
option with 10 folds.
Procedure for percentage split:

3. In which we selected Linear Regression algorithm & click on start option with percentage split
option with 66% split.
C. Explore Simple linear regression technique that only looks at onevariable
Procedure:
3. In which we selected Simple Linear Regression algorithm & click on start option with use
training set option with one variable (MYCT).
Data Mining
Credit Risk Assesment – The German Credit Data
Task 1:- list all the categorical (or nominal) attributes and the real-valued attributes separately.
Procedure:-
1) Insert the data into the excel sheet and save the file is “.CSV”.
2) Click on weka executable jar file.
3) A window can open that window contains the buttons that are “Explorer, Experimenter, Knowledge flow,
simple CLI”.
4) Click the “Explorer” button.
5) A new window is open and click the “preprocess” tab and open a “german.arff ” file.
6) You can see the list of categorical and nominal attributes can be dispayed.
Output:-
Title: German Credit data
Number of Instances: 1000
The categorical attributes are 13 that are as follows.
For algorithms that need numerical attributes, Strathclyde University produced the file "german.data-numeric". This
file has been edited and several indicator variables added to make it suitable for algorithms which cannot cope with
categorical variables.Severalattributes that are ordered categorical (such as attribute 17) havebeen coded as integer.
This was the form used by StatLog.
6. Number of Attributes german: 20 (7 numerical, 13 categorical)

Number of Attributes german.numer: 24 (24 numerical)
7. Attribute description for german
Attribute 1: (qualitative)
Status of existing checking account
A11 : ... < 0 DM
A12 : 0 <= ... < 200 DM
A13 : ... >= 200 DM /
salary assignments for at least 1 year
A14 : no checking account
Attribute 2: (numerical)
Duration in month
Credit history
A30 : no credits taken/
all credits paid back duly
A31 : all credits at this bank paid back duly
A32 : existing credits paid back duly till now
A33 : delay in paying off in the past
A34 : critical account/
other credits existing (not at this bank)
Purpose
A40 : car (new)
A41 : car (used)
A42 : furniture/equipment
A43 : radio/television
A44 : domestic appliances
A45 : repairs
A46 : education
A47 : (vacation - does not exist?)
A48 : retraining
A49 : business
A410 : others
Credit amount
Attibute 6: (qualitative)
Savings account/bonds
A61 : ... < 100 DM
A62 : 100 <= ... < 500 DM
A63 : 500 <= ... < 1000 DM
A64 : .. >= 1000 DM
A65 : unknown/ no savings account
Present employment since
A71 : unemployed
A72 : ... < 1 year
A73 : 1 <= ... < 4 years
A74 : 4 <= ... < 7 years
A75 : .. >= 7 years
Installment rate in percentage of disposable income
Personal status and sex
A91 : male : divorced/separated
A92 : female : divorced/separated/married
A93 : male : single
A94 : male : married/widowed
A95 : female : single

Other debtors / guarantors
A101 : none
A102 : co-applicant
A103 : guarantor
Present residence since

Property
A121 : real estate
A122 : if not A121 : building society savings agreement/
life insurance
A123 : if not A121/A122 : car or other, not in attribute 6
A124 : unknown / no property

Age in years

Other installment plans
A141 : bank
A142 : stores
A143 : none

Housing
A151 : rent
A152 : own
A153 : for free

Number of existing credits at this bank

Job
A171 : unemployed/ unskilled - non-resident
A172 : unskilled - resident
A173 : skilled employee / official
A174 : management/ self-employed/
highly qualified employee/ officer

Number of people being liable to provide maintenance for

Telephone
A191 : none
A192 : yes, registered under the customers name
foreign worker
A201 : yes
A202 : no
Task 2:- what attributes do you think might be crucial in making the credit assessment ? come up with some
simple rules in plain English using your selected attributes
Procedure:-
3) A window can open that window contains the buttons that are ”Explorer, Experimenter, Knowledge flow,
simple CLI”.
6) Choose the credit history has the crucial attributes are 293.
Output:-
Task 3:- one type of model that you can create is a Decision tree – train a Decision tree using the complete dataset
as the training data. Report the model obtained after training.
Procedure:-
simple CLI”.
6) Next click the “classify” menu on the top.
7) Click the “choose” and list of “trees and rules” can be displayed.
8) Select the “trees” under click the “J48”. And click “start”.
9) Result list can be displayed and right click on “J48” and choose the and select the “visualize tree”. See the
decision tree.
Output:-
Task 4:-
Suppose you use your above model trained on the complete dataset, and classify credit good/bad for each of the
examples in the dataset. What % of examples can you classify correctly? (this is also called testing on the training
set) why do you think you cannot get 100% training accuracy?
Procedure:-
3) A window can open that window contains the for buttons that are ”Explorer, Experimenter, Knowledge
flow, simple CLI”.
8) Select the “trees” under click the “J48”. And choose the radio button is “using training dataset”. And click
“start”. See the correctly classified instances.
Output:-
=== Run information ===
Scheme: weka.classifiers.trees.J48 -C 0.25 -M 2

Relation: credit-g-weka.filters.unsupervised.attribute.Remove-R3
Instances: 1000
Attributes: 20
checking_status
duration
purpose
credit_amount
savings_status
employment
installment_commitment
personal_status
other_parties
residence_since
property_magnitude
age
other_payment_plans
housing
existing_credits
job
num_dependents
own_telephone
foreign_worker
class
Test mode: evaluate on training data
=== Classifier model (full training set) ===
J48 pruned tree

------------------
checking_status = <0
| foreign_worker = yes
| | duration <= 11
| | | existing_credits <= 1
| | | | property_magnitude = real estate: good (8.0/1.0)
| | | | property_magnitude = life insurance
| | | | | own_telephone = yes: good (4.0)
| | | | | own_telephone = none: bad (2.0)
| | | | property_magnitude = no known property: bad (3.0)
| | | | property_magnitude = car: good (2.0/1.0)
| | | existing_credits > 1: good (14.0)
| | duration > 11
| | | job = skilled
| | | | other_parties = none
| | | | | savings_status = no known savings
| | | | | | existing_credits <= 1
| | | | | | | own_telephone = yes: good (4.0/1.0)
| | | | | | | own_telephone = none: bad (10.0/1.0)
| | | | | | existing_credits > 1: good (2.0)
| | | | | savings_status = <100: bad (98.0/30.0)
| | | | | savings_status = 500<=X<1000: good (5.0/2.0)
| | | | | savings_status = >=1000: good (4.0)
| | | | | savings_status = 100<=X<500
| | | | | | property_magnitude = real estate: good (1.0)
| | | | | | property_magnitude = life insurance: bad (3.0)
| | | | | | property_magnitude = no known property: good (0.0)
| | | | | | property_magnitude = car: good (2.0)
| | | | other_parties = guarantor
| | | | | duration <= 45: good (10.0/1.0)
| | | | | duration > 45: bad (2.0)
| | | | other_parties = co applicant: bad (7.0/1.0)
| | | job = unskilled resident
| | | | purpose = radio/tv
| | | | | existing_credits <= 1: bad (10.0/3.0)
| | | | | existing_credits > 1: good (2.0)
| | | | purpose = education: bad (1.0)
| | | | purpose = furniture/equipment
| | | | | employment = >=7: good (2.0)
| | | | | employment = 1<=X<4: good (4.0)
| | | | | employment = unemployed: good (0.0)
| | | | | employment = <1: bad (3.0)
| | | | purpose = new car
| | | | | own_telephone = none: bad (10.0/2.0)
| | | | purpose = used car: bad (1.0)
| | | | purpose = business: good (3.0)
| | | | purpose = domestic appliance: bad (1.0)
| | | | purpose = repairs: bad (1.0)
| | | | purpose = other: good (1.0)
| | | | purpose = retraining: good (1.0)
| | | job = high qualif/self emp/mgmt: good (30.0/8.0)
| | | job = unemp/unskilled non res: bad (5.0/1.0)
| foreign_worker = no: good (15.0/2.0)
checking_status = 0<=X<200
| credit_amount <= 9857
| | savings_status = no known savings: good (41.0/5.0)
| | savings_status = <100
| | | other_parties = none
| | | | duration <= 42
| | | | | personal_status = male single: good (52.0/15.0)
| | | | | personal_status = female div/dep/mar
| | | | | | purpose = radio/tv: good (8.0/2.0)
| | | | | | purpose = education: good (4.0/2.0)
| | | | | | purpose = furniture/equipment
| | | | | | | duration <= 10: bad (3.0)
| | | | | | | duration > 10
| | | | | | | | duration <= 21: good (6.0/1.0)
| | | | | | | | duration > 21: bad (2.0)
| | | | | | purpose = new car: bad (5.0/1.0)
| | | | | | purpose = used car: bad (1.0)
| | | | | | purpose = business
| | | | | | | residence_since <= 2: good (3.0)
| | | | | | | residence_since > 2: bad (2.0)
| | | | | | purpose = domestic appliance: good (0.0)
| | | | | | purpose = repairs: good (1.0)
| | | | | | purpose = other: good (0.0)
| | | | | | purpose = retraining: good (0.0)
| | | | | personal_status = male div/sep: bad (8.0/2.0)
| | | | | personal_status = male mar/wid
| | | | | | duration <= 10: good (6.0)
| | | | | | duration > 10: bad (10.0/3.0)
| | | | duration > 42: bad (7.0)
| | | other_parties = guarantor
| | | | purpose = radio/tv: good (18.0/1.0)
| | | | purpose = education: good (0.0)
| | | | purpose = furniture/equipment: good (0.0)
| | | | purpose = new car: bad (2.0)
| | | | purpose = used car: good (0.0)
| | | | purpose = domestic appliance: good (0.0)
| | | | purpose = repairs: good (0.0)
| | | other_parties = co applicant: good (2.0)
| | savings_status = 500<=X<1000: good (11.0/3.0)
| | savings_status = >=1000
| | | duration <= 10: bad (2.0)
| | | duration > 10: good (11.0/1.0)
| | savings_status = 100<=X<500
| | | purpose = radio/tv: bad (8.0/2.0)
| | | purpose = education: good (0.0)
| | | purpose = furniture/equipment: bad (4.0/1.0)
| | | purpose = new car: bad (15.0/5.0)
| | | purpose = used car: good (3.0)
| | | purpose = business
| | | | housing = own: good (6.0)
| | | | housing = for free: bad (1.0)
| | | | housing = rent
| | | | | existing_credits <= 1: good (2.0)
| | | | | existing_credits > 1: bad (2.0)
| | | purpose = domestic appliance: good (0.0)
| | | purpose = repairs: good (2.0)
| | | purpose = other: good (1.0)
| | | purpose = retraining: good (0.0)
| credit_amount > 9857: bad (20.0/3.0)
checking_status = no checking: good (394.0/46.0)
checking_status = >=200: good (63.0/14.0)
Number of Leaves : 87
Size of the tree : 119
Time taken to build model: 0.02 seconds
=== Evaluation on training set ===

=== Summary ===
Correctly Classified Instances 839 83.9 %

Incorrectly Classified Instances 161 16.1 %
Kappa statistic 0.5971
Mean absolute error 0.2508
Root mean squared error 0.3541
Relative absolute error 59.6831 %
Root relative squared error 77.2695 %
Total Number of Instances 1000
=== Detailed Accuracy By Class ===
TP Rate FP Rate Precision Recall F-Measure ROC Area Class

0.921 0.353 0.859 0.921 0.889 0.848 good
0.647 0.079 0.779 0.647 0.707 0.848 bad
Weighted Avg. 0.839 0.271 0.835 0.839 0.834 0.848
=== Confusion Matrix ===
a b <-- classified as
645 55 | a = good
106 194 | b = bad
Task 5:-
Is testing on the training set as you did above a good idea? Why or why not?
Description:-
⮚ Performance on the training set is definitely not a good indicator of performance on an independent test
data.
⮚ For classification problem it is naturally to measures a classifiers performance in terms of the error rate.
The classifier predicts the class of each instance if it is correct that is counted as a success if not it is an
error.
⮚ The error rate on the training data is called the “ resubstitution error”, because it is calculated by
resubstituting the training instances into a classifier that was constructed for them.
⮚ None of the data may be used to determine an estimate of the future error rate. In such situations uses the
training data, validation data, and test data.
⮚ The training data is used by one or more learning methods to come up with classifiers.
⮚ Of course, what are interested in is the likely future performance on new data, not the past performance on
old data. We already know the classification of each instance in the training set, which after all is why we
can use it for training.
⮚ Data are randomly partitioned into two independent sets, a training set and test set.
⮚ Typically two-third of the data are allocated to the training set, and the remaining one-third allocated to the
test set.
Task 6:-
One approach for solving the problem encountered in the previous questions is using cross-validation? Describe
what is cross-validation briefly. Train a decision tree again using cross-validation and report your results. Does your
accuracy increase or decrease? Why?
Procedure:-
8) Select the “trees” under click the “J48”. And check the radio button is cross-validation. And click “start”.
See the correctly classified instances.
In cross-validation you decided on a fixed number of folds or partitions of the data. Two-third for training and one-
third for testing and repeat procedure three times so that in the end, every instance has been used exactly once for
testing is called stratified cross-fold validation.In training dataset compared with cross-validation the accuracy is
decreases.
Output:-

Relation: credit-g-weka.filters.unsupervised.attribute.Remove-R3
Instances: 1000
Attributes: 20
checking_status
duration
purpose
credit_amount
savings_status
employment
personal_status
other_parties
residence_since
property_magnitude
age
other_payment_plans
housing
existing_credits
job
num_dependents
own_telephone
foreign_worker
class
Test mode: 10-fold cross-validation
J48 pruned tree

------------------
| foreign_worker = yes
| | duration <= 11
| | | existing_credits <= 1
| | | | property_magnitude = real estate: good (8.0/1.0)
| | | | property_magnitude = life insurance
| | | | | own_telephone = none: bad (2.0)
| | | | property_magnitude = no known property: bad (3.0)
| | | | property_magnitude = car: good (2.0/1.0)
| | | existing_credits > 1: good (14.0)
| | duration > 11
| | | job = skilled
| | | | other_parties = none
| | | | | | existing_credits <= 1
| | | | | | | own_telephone = yes: good (4.0/1.0)
| | | | | | | own_telephone = none: bad (10.0/1.0)
| | | | | | existing_credits > 1: good (2.0)
| | | | | savings_status = <100: bad (98.0/30.0)
| | | | | savings_status = 100<=X<500
| | | | | | property_magnitude = real estate: good (1.0)
| | | | | | property_magnitude = life insurance: bad (3.0)
| | | | | | property_magnitude = no known property: good (0.0)
| | | | | | property_magnitude = car: good (2.0)
| | | | other_parties = guarantor
| | | | | duration <= 45: good (10.0/1.0)
| | | | | duration > 45: bad (2.0)
| | | | other_parties = co applicant: bad (7.0/1.0)
| | | job = unskilled resident
| | | | purpose = radio/tv
| | | | | existing_credits <= 1: bad (10.0/3.0)
| | | | | existing_credits > 1: good (2.0)
| | | | purpose = education: bad (1.0)
| | | | | employment = >=7: good (2.0)
| | | | | employment = unemployed: good (0.0)
| | | | | employment = <1: bad (3.0)
| | | | | own_telephone = none: bad (10.0/2.0)
| | | | purpose = used car: bad (1.0)
| | | | purpose = domestic appliance: bad (1.0)
| | | | purpose = repairs: bad (1.0)
| | | job = high qualif/self emp/mgmt: good (30.0/8.0)
| | | job = unemp/unskilled non res: bad (5.0/1.0)
| foreign_worker = no: good (15.0/2.0)
| | | | duration <= 42
| | | | | personal_status = male single: good (52.0/15.0)
| | | | | personal_status = female div/dep/mar
| | | | | | purpose = radio/tv: good (8.0/2.0)
| | | | | | purpose = education: good (4.0/2.0)
| | | | | | purpose = furniture/equipment
| | | | | | | duration <= 10: bad (3.0)
| | | | | | | duration > 10
| | | | | | | | duration <= 21: good (6.0/1.0)
| | | | | | | | duration > 21: bad (2.0)
| | | | | | purpose = new car: bad (5.0/1.0)
| | | | | | purpose = used car: bad (1.0)
| | | | | | purpose = business
| | | | | | | residence_since <= 2: good (3.0)
| | | | | | | residence_since > 2: bad (2.0)
| | | | | | purpose = domestic appliance: good (0.0)
| | | | | | purpose = repairs: good (1.0)
| | | | | | purpose = other: good (0.0)
| | | | | | purpose = retraining: good (0.0)
| | | | | personal_status = male div/sep: bad (8.0/2.0)
| | | | | personal_status = male mar/wid
| | | | | | duration <= 10: good (6.0)
| | | | | | duration > 10: bad (10.0/3.0)
| | | | duration > 42: bad (7.0)
| | | other_parties = guarantor
| | | | purpose = education: good (0.0)
| | | | purpose = furniture/equipment: good (0.0)
| | | | purpose = new car: bad (2.0)
| | | | purpose = used car: good (0.0)
| | | | purpose = repairs: good (0.0)
| | | other_parties = co applicant: good (2.0)
| | savings_status = >=1000
| | | duration <= 10: bad (2.0)
| | | duration > 10: good (11.0/1.0)
| | | purpose = new car: bad (15.0/5.0)
checking_status = >=200: good (63.0/14.0)
=== Stratified cross-validation ===

=== Summary ===


0.83 0.593 0.765 0.83 0.796 0.634 good
0.407 0.17 0.506 0.407 0.451 0.634 bad
Weighted Avg. 0.703 0.466 0.688 0.703 0.693 0.634
581 119 | a = good
178 122 | b = bad
The accuracy is decreases compared with the use training dataset.
Task 7:-
Check to see if the data shows a bias against “foreign workers” (attribute 20) or “personal-status” (attribute 9(. One
way to do this (perhaps rather simple minded) is to remove these attributes from the dataset and see if the decision
tree created in those cases is significantly different from the full dataset case which you have already done. To
remove an attribute you can use the preprocess tab in weka in GUI Explorer. Did removing these attributes
Procedure:-
4) Click the “Explorer” button. A new window is open and click the “preprocess” tab and open a “german.arff
” file.
5) Next we can remove the foreign workers and personal-status attributes.
8) Select the “trees” under click the “J48”. And next right click on j48tree and click the “visualize tree”
Task 8:-
Another question might be do you really need to input so many attributes to get good results? May be only a few
would do. For example you could try just having attributes 2,3,5,7,10,17 (and 21 the class attribute naturally)). Try
out some combinations(you had removed two attributes in problem 7. Remember to reload the arff data file to get all
the attributes inintially before you start selecting the ones you want)
Procedure:-
simple CLI”.
6) Choose “visualize all” button and see the result.
Output:-
Task 9:-
Sometimes the cost of rejecting an applicant who actually has a good credit (case 1) might be higher than accepting
an applicant who had bad credit (case 2). Instead of counting the misclassifications equally in both cases, give a
higher cost to the first case (say cost 5) and lower cost to the second case. You can do this by using a cost matrix in
weka. Train your decision tree against and report the decision tree and cross validation results. Are they significantly
different from results obtained in problem 6 (using equal cost)
Procedure:-
9) Choose on result list and right click that one and click visualize “cost curve” is good can select.
Output:-
Using cross-validation dataset results


Relation: credit-g-weka.filters.unsupervised.attribute.Remove-R9,20
Instances: 1000
Attributes: 19
checking_status
duration
credit_history
purpose
credit_amount
savings_status
employment
other_parties
residence_since
property_magnitude
age
other_payment_plans
housing
existing_credits
job
num_dependents
own_telephone
class
Test mode: 10-fold cross-validation
J48 pruned tree

------------------
| duration <= 11
| | existing_credits <= 1
| | | property_magnitude = real estate: good (9.0/1.0)
| | | property_magnitude = life insurance
| | | | own_telephone = yes: good (4.0)
| | | | own_telephone = none: bad (2.0)
| | | property_magnitude = no known property: bad (3.0)
| | | property_magnitude = car: good (2.0/1.0)
| | existing_credits > 1: good (19.0)
| duration > 11
| | job = skilled
| | | | duration <= 30
| | | | | | own_telephone = yes: good (6.0/1.0)
| | | | | | own_telephone = none
| | | | | | | installment_commitment <= 3: good (3.0/1.0)
| | | | | | | installment_commitment > 3: bad (7.0)
| | | | | savings_status = <100
| | | | | | credit_history = critical/other existing credit: good (14.0/4.0)
| | | | | | credit_history = existing paid
| | | | | | | own_telephone = yes: bad (5.0)
| | | | | | | own_telephone = none
| | | | | | | | employment = >=7: good (2.0)
| | | | | | | | employment = 1<=X<4
| | | | | | | | | age <= 26: bad (7.0/1.0)
| | | | | | | | | age > 26: good (7.0/1.0)
| | | | | | | | employment = 4<=X<7: bad (5.0)
| | | | | | | | employment = unemployed: good (3.0/1.0)
| | | | | | | | employment = <1
| | | | | | | | | property_magnitude = real estate: good (2.0)
| | | | | | | | | property_magnitude = life insurance: bad (4.0)
| | | | | | | | | property_magnitude = no known property: good (1.0)
| | | | | | | | | property_magnitude = car: good (3.0)
| | | | | | credit_history = delayed previously: bad (4.0)
| | | | | | credit_history = no credits/all paid: bad (8.0/1.0)
| | | | | | credit_history = all paid: bad (6.0)
| | | | | savings_status = 100<=X<500
| | | | | | credit_history = critical/other existing credit: good (2.0)
| | | | | | credit_history = existing paid: bad (3.0)
| | | | | | credit_history = delayed previously: good (0.0)
| | | | | | credit_history = no credits/all paid: good (0.0)
| | | | | | credit_history = all paid: good (1.0)
| | | | duration > 30: bad (30.0/3.0)
| | | other_parties = guarantor: good (14.0/4.0)
| | | other_parties = co applicant: bad (7.0/1.0)
| | job = unskilled resident
| | | property_magnitude = real estate
| | | | existing_credits <= 1
| | | | | num_dependents <= 1
| | | | | | installment_commitment <= 2: good (3.0)
| | | | | | installment_commitment > 2: bad (10.0/4.0)
| | | | | num_dependents > 1: bad (2.0)
| | | | existing_credits > 1: good (3.0)
| | | | duration <= 18: good (9.0)
| | | | duration > 18: bad (3.0/1.0)
| | | property_magnitude = car: bad (12.0/5.0)
| | job = high qualif/self emp/mgmt: good (31.0/9.0)
| | job = unemp/unskilled non res: bad (5.0/1.0)
| | | duration <= 42
| | | | purpose = education
| | | | | age <= 33: good (2.0)
| | | | | age > 33: bad (3.0/1.0)
| | | | | other_payment_plans = none
| | | | | | housing = own: bad (14.0/5.0)
| | | | | | housing = for free: bad (0.0)
| | | | | | housing = rent: good (5.0/1.0)
| | | | | other_payment_plans = bank: good (2.0/1.0)
| | | | | other_payment_plans = stores: good (2.0)
| | | | | employment = >=7: bad (5.0)
| | | | | employment = 1<=X<4: good (5.0/2.0)
| | | | | employment = unemployed
| | | | | | installment_commitment > 3: bad (3.0)
| | | | | employment = <1: bad (7.0/2.0)
| | | | purpose = used car
| | | | | residence_since <= 3: good (6.0)
| | | | | residence_since > 3: bad (3.0/1.0)
| | | | purpose = business
| | | | | residence_since <= 3: good (10.0/2.0)
| | | | | residence_since > 3: bad (5.0)
| | | | purpose = repairs
| | | | | installment_commitment <= 3: good (3.0)
| | | | | installment_commitment > 3: bad (3.0/1.0)
| | | duration > 42: bad (7.0)
| | savings_status = >=1000: good (13.0/3.0)
| | | purpose = new car
| | | | property_magnitude = real estate: bad (0.0)
| | | | property_magnitude = life insurance: bad (6.0)
| | | | property_magnitude = no known property: good (2.0/1.0)
| | | | property_magnitude = car
checking_status = >=200
| property_magnitude = real estate
| | installment_commitment <= 3: good (15.0/3.0)
| | installment_commitment > 3: bad (6.0/1.0)
| property_magnitude = life insurance: good (12.0)
| property_magnitude = no known property
| | num_dependents <= 1: good (7.0/1.0)
| | num_dependents > 1: bad (2.0)
| property_magnitude = car: good (21.0/3.0)
=== Summary ===

0.839 0.57 0.774 0.839 0.805 0.66 good
0.43 0.161 0.533 0.43 0.476 0.66 bad
Weighted Avg. 0.716 0.447 0.702 0.716 0.706 0.66

587 113 | a = good
171 129 | b = bad
Cost curve:-
Using training dataset results

Relation: credit-g-weka.filters.unsupervised.attribute.Remove-R9,20
Instances: 1000
Attributes: 19
checking_status
duration
credit_history
purpose
credit_amount
savings_status
employment
other_parties
residence_since
property_magnitude
age
other_payment_plans
housing
existing_credits
job
num_dependents
own_telephone
class
J48 pruned tree

------------------
| duration <= 11
| | existing_credits <= 1
| | | property_magnitude = real estate: good (9.0/1.0)
| | | | own_telephone = yes: good (4.0)
| | | | own_telephone = none: bad (2.0)
| | | property_magnitude = car: good (2.0/1.0)
| | existing_credits > 1: good (19.0)
| duration > 11
| | job = skilled
| | | | duration <= 30
| | | | | | own_telephone = yes: good (6.0/1.0)
| | | | | | own_telephone = none
| | | | | | | installment_commitment <= 3: good (3.0/1.0)
| | | | | | | installment_commitment > 3: bad (7.0)
| | | | | savings_status = <100
| | | | | | credit_history = critical/other existing credit: good (14.0/4.0)
| | | | | | credit_history = existing paid
| | | | | | | own_telephone = yes: bad (5.0)
| | | | | | | own_telephone = none
| | | | | | | | employment = >=7: good (2.0)
| | | | | | | | employment = 1<=X<4
| | | | | | | | | age <= 26: bad (7.0/1.0)
| | | | | | | | | age > 26: good (7.0/1.0)
| | | | | | | | employment = 4<=X<7: bad (5.0)
| | | | | | | | employment = unemployed: good (3.0/1.0)
| | | | | | | | employment = <1
| | | | | | | | | property_magnitude = real estate: good (2.0)
| | | | | | | | | property_magnitude = life insurance: bad (4.0)
| | | | | | | | | property_magnitude = no known property: good (1.0)
| | | | | | | | | property_magnitude = car: good (3.0)
| | | | | | credit_history = delayed previously: bad (4.0)
| | | | | | credit_history = no credits/all paid: bad (8.0/1.0)
| | | | | | credit_history = all paid: bad (6.0)
| | | | | savings_status = 100<=X<500
| | | | | | credit_history = critical/other existing credit: good (2.0)
| | | | | | credit_history = existing paid: bad (3.0)
| | | | | | credit_history = delayed previously: good (0.0)
| | | | | | credit_history = no credits/all paid: good (0.0)
| | | | | | credit_history = all paid: good (1.0)
| | | | duration > 30: bad (30.0/3.0)
| | | other_parties = guarantor: good (14.0/4.0)
| | | other_parties = co applicant: bad (7.0/1.0)
| | job = unskilled resident
| | | property_magnitude = real estate
| | | | existing_credits <= 1
| | | | | num_dependents <= 1
| | | | | | installment_commitment > 2: bad (10.0/4.0)
| | | | | num_dependents > 1: bad (2.0)
| | | | existing_credits > 1: good (3.0)
| | | | duration <= 18: good (9.0)
| | | | duration > 18: bad (3.0/1.0)
| | | property_magnitude = car: bad (12.0/5.0)
| | job = high qualif/self emp/mgmt: good (31.0/9.0)
| | job = unemp/unskilled non res: bad (5.0/1.0)
| | | duration <= 42
| | | | purpose = education
| | | | | age <= 33: good (2.0)
| | | | | age > 33: bad (3.0/1.0)
| | | | | other_payment_plans = none
| | | | | | housing = own: bad (14.0/5.0)
| | | | | | housing = for free: bad (0.0)
| | | | | | housing = rent: good (5.0/1.0)
| | | | | other_payment_plans = bank: good (2.0/1.0)
| | | | | other_payment_plans = stores: good (2.0)
| | | | | employment = >=7: bad (5.0)
| | | | | employment = unemployed
| | | | | | installment_commitment > 3: bad (3.0)
| | | | | employment = <1: bad (7.0/2.0)
| | | | purpose = used car
| | | | purpose = business
| | | | | residence_since <= 3: good (10.0/2.0)
| | | | | residence_since > 3: bad (5.0)
| | | | purpose = repairs
| | | | | installment_commitment <= 3: good (3.0)
| | | | | installment_commitment > 3: bad (3.0/1.0)
| | | duration > 42: bad (7.0)
| | savings_status = >=1000: good (13.0/3.0)
| | | purpose = new car
| | | | property_magnitude = real estate: bad (0.0)
| | | | property_magnitude = life insurance: bad (6.0)
| | | | property_magnitude = no known property: good (2.0/1.0)
| | | | property_magnitude = car
checking_status = >=200
| property_magnitude = real estate
| | installment_commitment <= 3: good (15.0/3.0)
| | installment_commitment > 3: bad (6.0/1.0)
| property_magnitude = life insurance: good (12.0)
| property_magnitude = no known property
| | num_dependents <= 1: good (7.0/1.0)
| | num_dependents > 1: bad (2.0)
| property_magnitude = car: good (21.0/3.0)
=== Summary ===
0.95 0.347 0.865 0.95 0.905 0.869 good
0.653 0.05 0.848 0.653 0.738 0.869 bad
Weighted Avg. 0.861 0.258 0.86 0.861 0.855 0.869
665 35 | a = good
104 196 | b = bad
Cost curve:-
Task 10:-
Do you think it is a good idea to prefer simple decision tree instead of having long complex decision trees? How
does the complexity of a decision tree relate to the bias of the model?
Description:-
⮚ The computational complexity of the decision tree induction is O(n) stands for a quantity that grows at
most linearly with n, O(n2) grows at most quadratic ally with n.
⮚ Suppose the training data contains n instances and m attributes. We need to make some assumption about
the size of the tree, and we will assume that its depth is on the order of log n, that is O(log n).
⮚ The computational cost of building the tree in the first place is O(mn log n).
⮚ Because there are log n different depths in the tree, the amount of work for this on attribute is O(n log n) .
at each node all attributes are considered so the total amount of work is O(mn log n).
⮚ The initial sort takes O(n log n) operations for each of up to m attributes thus the preceding complexity
figure is unchanged.
⮚ The complexity of sub-tree replacement is O(n).
⮚ Finally sub-tree lifting has a basic complexity equal to sub-tree replacement. But there is an added cost
because instances need to be reclassified during the lifting operation. During the whole process each
instance may have to be reclassified at every node between its leaf and the root, I.e, as many as O(log n)
times. That makes the total number of reclassification O(n log n )
⮚ And reclassification is not a single operation one that occurs near the root will take O(log n) operations,
and one of the average depth will take half of this. Thus the total complexity of sub-tree is as follows.
⮚ O(n(log n)2)\taking into account all these operations, the full complexity of decision tree induction is
⮚ O(mn log n)+O(n(log n)2 ).
Task 11:-
You can make your Decision tree simpler by pruning the nodes. One approach is to use Reduce error pruning-
Explain this idea briefly. Try reduce error pruning for training your Decision tree using cross-validation (you can do
this in weka) and report the Decision tree you obtain? Also, report your accuracy using the pruned model. Does your
accuracy increases?
Description:-
We need to estimate the error at internal nodes as well as at leaf nodes. If we had such an estimate, it would be clear
whether to replace, or raise, a particular sub-tree simply by comparing the estimated error of the sub-tree with that of
its proposed replacement. Before estimating the error for sub-tree proposed for raising.
One Way of coming up with an error estimate is the standard verification technique is hold back some of the data
originally given and use it as an independent test set to estimate the error at each node. This is called reduced-error
pruning.
Procedure:-
Output:-
0.956 0.38 0.854 0.956 0.902 0.857 good
0.62 0.044 0.857 0.62 0.72 0.857 bad
Weighted Avg. 0.855 0.279 0.855 0.855 0.847 0.857
0.843 0.61 0.763 0.843 0.801 0.641 good
0.39 0.157 0.515 0.39 0.444 0.641 bad
Weighted Avg. 0.707 0.474 0.689 0.707 0.694 0.641
Task 12:-
How can you convert a Decision tree into “if-then-else-rules”. Make up your own small Decision tree consisting of
2-3 level s and convert it into a set of rules. There also exist different classifiers that output the model in the form of
rules- one such classifier in weka is rules. PART, train this model and report the set of rules obtained. Sometimes
just one attribute can be good enough in making the decision, yes, just one. Can you predict what attribute that might
be in this dataset? OneR classifier uses a single attribute to make decisions (it chooses the attribute based on
minimum error). Report the rule obtained by training a OneR classifier. Rank the performance of j48,PART,and
OneR.
Procedure:-
8) Select the “rules” under click the “PART and OneR”. And choose the radio button is “using training
dataset”. And click “start”. See the correctly classified instances.
Output:-
PART Rule
Scheme: weka.classifiers.rules.PART -M 2 -C 0.25 -Q 1
Relation: credit-g
Instances: 1000
Attributes: 21
checking_status
duration
credit_history
purpose
credit_amount
savings_status
employment
personal_status
other_parties
residence_since
property_magnitude
age
other_payment_plans
housing
existing_credits
job
num_dependents
own_telephone
foreign_worker
class
PART decision list
------------------
checking_status = no checking AND
other_payment_plans = none AND
credit_history = critical/other existing credit: good (134.0/3.0)
existing_credits <= 1 AND
purpose = radio/tv: good (49.0/2.0)
foreign_worker = yes AND
employment = 4<=X<7: good (35.0/2.0)
foreign_worker = no AND
personal_status = male single: good (21.0)
purpose = used car AND
other_payment_plans = none: good (23.0)
duration <= 15 AND
other_parties = guarantor: good (22.0/1.0)
duration <= 11 AND
credit_history = critical/other existing credit: good (29.0/3.0)
checking_status = >=200 AND
num_dependents <= 1 AND
property_magnitude = car: good (20.0/3.0)
property_magnitude = real estate AND
age > 23: good (25.0)
savings_status = >=1000 AND
property_magnitude = real estate: good (10.0)
savings_status = 500<=X<1000 AND
employment = >=7: good (13.0/1.0)
credit_history = no credits/all paid AND
housing = rent: bad (9.0)

savings_status = no known savings AND
checking_status = 0<=X<200 AND
existing_credits > 1: good (9.0)
property_magnitude = life insurance: good (9.0)
installment_commitment <= 2 AND
other_parties = co applicant AND
existing_credits > 1: bad (5.0)
credit_history = delayed previously AND
existing_credits > 1 AND
residence_since > 1: good (14.0/3.0)
existing_credits <= 1: good (9.0)
duration > 30 AND
savings_status = 100<=X<500: bad (13.0/3.0)
credit_history = all paid AND
other_parties = none AND
other_payment_plans = bank: bad (16.0/5.0)
duration > 30 AND
num_dependents > 1: good (5.0)
duration > 30 AND
credit_history = delayed previously: bad (9.0)
duration > 42 AND
savings_status = <100 AND
residence_since > 1: bad (28.0/3.0)
credit_amount <= 8133 AND
existing_credits > 1: good (11.0)

credit_amount > 8133: bad (8.0/1.0)
employment = 1<=X<4: good (7.0)
purpose = used car: good (16.0/3.0)
purpose = furniture/equipment AND
other_payment_plans = stores: good (8.0)
credit_history = all paid AND
other_payment_plans = none: bad (10.0)
purpose = business AND
residence_since <= 1: good (9.0)
other_payment_plans = stores AND
purpose = radio/tv AND
personal_status = male single: bad (6.0/1.0)
employment = >=7 AND
num_dependents <= 1: good (20.0/1.0)
own_telephone = yes: good (19.0/3.0)
personal_status = male single: good (11.0/1.0)
checking_status = 0<=X<200 AND
employment = 4<=X<7 AND
personal_status = male single AND
residence_since > 2: good (9.0)
purpose = other: good (5.0/1.0)

credit_history = existing paid AND
residence_since > 1 AND
housing = rent AND
installment_commitment <= 1: good (9.0)
housing = rent AND
purpose = new car: bad (13.0/2.0)
other_payment_plans = stores AND
property_magnitude = life insurance: bad (4.0/1.0)
other_payment_plans = bank AND
housing = rent: bad (7.0/1.0)
installment_commitment > 3 AND
existing_credits <= 1 AND
purpose = new car: bad (17.0/5.0)
job = skilled AND
credit_amount > 1283: good (5.0)
checking_status = >=200: bad (9.0/1.0)
duration <= 15 AND
property_magnitude = real estate AND
existing_credits <= 1: good (22.0/5.0)
num_dependents > 1 AND
own_telephone = yes: good (14.0/2.0)
purpose = repairs AND

property_magnitude = car: good (5.0/1.0)
purpose = education AND
checking_status = <0: bad (9.0/1.0)
purpose = repairs: bad (7.0/1.0)
checking_status = 0<=X<200: good (5.0)
personal_status = male single: bad (3.0)
purpose = business AND
age > 27: bad (14.0/3.0)
purpose = new car AND
other_payment_plans = bank: bad (13.0/2.0)
other_parties = guarantor AND
existing_credits <= 1: good (5.0/1.0)
other_payment_plans = bank AND
num_dependents <= 1: good (6.0/1.0)
checking_status = <0 AND
job = skilled: bad (9.0/1.0)

credit_amount > 2051: good (6.0)
purpose = business: good (5.0)
housing = for free AND
residence_since > 2: good (12.0/3.0)
housing = own AND
checking_status = <0 AND
property_magnitude = car: bad (6.0/1.0)
existing_credits > 1 AND
credit_amount <= 2225: good (4.0)
job = unskilled resident: bad (6.0)
other_parties = co applicant AND
duration > 15: bad (4.0)
employment = unemployed AND
residence_since <= 2: bad (9.0)

property_magnitude = car AND
installment_commitment <= 2: good (5.0)
own_telephone = yes: bad (5.0/1.0)
own_telephone = none AND
duration > 27: bad (9.0/1.0)
personal_status = male single AND
residence_since > 3: good (6.0)
personal_status = female div/dep/mar AND
job = skilled AND
employment = 1<=X<4: bad (6.0/2.0)
housing = own AND
personal_status = male single: bad (5.0/1.0)
purpose = furniture/equipment: good (25.0/5.0)
personal_status = male mar/wid: bad (5.0)

own_telephone = none: good (8.0/1.0)
own_telephone = yes AND
age > 29: good (5.0/1.0)
age <= 45: good (4.0)
housing = own AND
personal_status = female div/dep/mar AND
installment_commitment > 1: bad (10.0)
housing = own AND
existing_credits <= 1: good (4.0)
housing = own AND
installment_commitment > 3 AND
savings_status = <100: bad (7.0/2.0)
checking_status = <0: bad (10.0/2.0)
: good (12.0/3.0)
Number of Rules : 78
=== Summary ===

0.941 0.167 0.929 0.941 0.935 0.949 good
0.833 0.059 0.859 0.833 0.846 0.949 bad
Weighted Avg. 0.909 0.134 0.908 0.909 0.909 0.949
659 41 | a = good
50 250 | b = bad
OneR Rule
Scheme: weka.classifiers.rules.OneR -B 6
Relation: credit-g
Instances: 1000
Attributes: 21
checking_status
duration
credit_history
purpose
credit_amount
savings_status
employment
personal_status
other_parties
residence_since
property_magnitude
age
other_payment_plans
housing
existing_credits
job
num_dependents
own_telephone
foreign_worker
class
credit_amount:
< 883.0 -> good
< 922.0 -> bad
< 938.0 -> good
< 979.5 -> bad
< 1206.5 -> good
< 1223.5 -> bad
< 1267.5 -> good
< 1286.0 -> bad
< 1821.5 -> good
< 1865.5 -> bad
< 3913.5 -> good

< 3969.0 -> bad
< 4049.5 -> good
< 4329.5 -> bad
< 4726.0 -> good
< 5024.0 -> bad
< 6322.5 -> good
< 6564.0 -> bad
< 6750.0 -> good
< 6917.5 -> bad
< 7760.5 -> good
< 8109.5 -> bad
< 9340.5 -> good
< 10331.5-> bad
< 11191.0-> good
>= 11191.0 -> bad
(742/1000 instances correct)
=== Summary ===

0.917 0.667 0.762 0.917 0.833 0.625 good
0.333 0.083 0.633 0.333 0.437 0.625 bad
Weighted Avg. 0.742 0.492 0.724 0.742 0.714 0.625
642 58 | a = good
200 100 | b = bad
Rank of the j48,PART,OneR is
Correctly classified instances in PART is:- 90.9%
Correctly classified instances in j48 is:- 83.3%
Correctly classified instances in OneR is:- 74.2%

DWM Lab Manual

Uploaded by

Copyright:

Available Formats

DWM Lab Manual

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

DWM Lab Manual

Uploaded by

Copyright:

Available Formats

(15A05610) DATA WAREHOUSING & MINING LABORATORY

1. Build Data Warehouse and Explore WEKA

(i). Identify source tables and populate sample data

After successful login, it will open new window as shown below.

Star Schema Model:

ETL Tool Implementation:

Procedure for OLAP Operations:

B. Explore WEKA Data Mining/Machine Learning Toolkit

(i). Downloading and/or installation of WEKA data mining toolkit

The buttons can be used to start the following applications:

Experimenter- An environment for performing experiments and conducting statistical tests

Searching and Evaluating

@ATTRIBUTE sepallength NUMERIC

The Data of the ARFF file looks like the following:

(v). Explore the available data sets in WEKA

(vii). Load each dataset and observe the following:

i. List the attribute names and they types

There are total 150 records (Instances) in dataset (IRIS.arff).

There is one class attribute which consists of 3 labels. They are:

iv. Plot Histogram

v. Determine the number of records for each class.

2. Perform data preprocessing tasks and Demonstrate performing association rule

Procedure for J48:

Procedure for Naïve-Bayes:

Procedure for ID3:

Procedure for J48:

Procedure for K-Nearest Neighbour (IBK):

4. Demonstrate performing clustering on data sets

B. Explore other clustering techniques available in Weka.

C. Explore visualization features of Weka to visualize the clusters. Derive interesting

5. Demonstrate performing Regression on data sets

Procedure for cross-validation:

Procedure for percentage split:

Title: German Credit data

Number of Instances: 1000

The categorical attributes are 13 that are as follows.

6. Number of Attributes german: 20 (7 numerical, 13 categorical)

7. Attribute description for german

Attribute 10: (qualitative)

Attribute 12: (qualitative)

Attribute 13: (numerical)

Attribute 14: (qualitative)

Attribute 15: (qualitative)

Attribute 16: (numerical)

Attribute 17: (qualitative)

Attribute 18: (numerical)

Attribute 19: (qualitative)

=== Run information ===

Scheme: weka.classifiers.trees.J48 -C 0.25 -M 2

=== Classifier model (full training set) ===

J48 pruned tree

Size of the tree : 119

Time taken to build model: 0.02 seconds

=== Evaluation on training set ===

Correctly Classified Instances 839 83.9 %

TP Rate FP Rate Precision Recall F-Measure ROC Area Class

=== Confusion Matrix ===

=== Run information ===

Scheme: weka.classifiers.trees.J48 -C 0.25 -M 2