Session 3: Data exploration

Aaron Ponti
The standard MATLAB variables are not specifically designed for statistical data.
Statistical data generally involves observations of multiple variables, with measurements of heterogeneous type and size.
Data may be numerical, categorical, or in the form of descriptive metadata. Fitting statistical data into basic MATLAB variables, and accessing it efficiently, can be
cumbersome.
Statistics Toolbox software offers two additional types of container variables
specifically designed for statistical data: categorical arrays and dataset arrays.

1
1.1

Data organization
Categorical arrays

Categorical data take on values from only a finite, discrete set of categories or levels.
Levels may have associated labels.
If no ordering is encoded in the levels, the data are nominal. Nominal labels typically indicate the type of an observation. Examples of nominal labels are { false, true
}, { male, female }, { Afghanistan, ..., Zimbabwe }. For nominal data, the numeric or
lexicographic order of the labels is irrelevant (Afghanistan is not considered to be less
than, equal to, or greater than Zimbabwe).
If an ordering is encoded in the levels (for example, if levels labeled {poor, satisfactory, outstanding} represent the result of an examination), the data are ordinal.
Labels for ordinal levels typically indicate the position or rank of an observation. (In
our example, an outstanding score is better than a poor one.)
For the sake of presenting the different categorical array types, we will
load the Iris flower data set (or Fishers Iris flower data set) introduced by
Sir Ronald Aylmer Fisher as an example of discriminant analysis in 1936 (see
http://en.wikipedia.org/wiki/Iris_flower_data_set). The dataset consists of 50 samples
from each of three species of Iris flowers (Iris setosa, Iris virginica and Iris versicolor).
Four features were measured from each sample, they are the length and the width of
sepal and petal, respectively (see Figure 1).
>> load fisheriris

% Fishers Iris data (1936)

loads the variables meas and species into the MATLAB workspace. The meas variable is a 150-by-4 numerical matrix, representing the 50 x 3 = 150 observations of 4 the
different measured variables. In statistics, observations are represented by the rows of
the matrix, while the measured variables (in our case sepal length, sepal width, petal
length, and petal width) are represented by the (four) columns.
1

DATA ORGANIZATION

Figure 1: Petals and sepals in Iris.

The 150-by-1 cell array species contains 50 x 3 = 150 labels ({setosa, setosa,
..., setosa, versicolor, versicolor, ..., versicolor, virginica, virginica, ..., virginica } ) that relates each of the measurement to one of the three Iris species.

>> meas( 1 : 10, : )

ans =
5.1000
4.9000
4.7000
4.6000
5.0000
5.4000
4.6000
5.0000
4.4000
4.9000

3.5000
3.0000
3.2000
3.1000
3.6000
3.9000
3.4000
3.4000
2.9000
3.1000

1.4000
1.4000
1.3000
1.5000
1.4000
1.7000
1.4000
1.5000
1.4000
1.5000

0.2000
0.2000
0.2000
0.2000
0.2000
0.4000
0.3000
0.2000
0.2000
0.1000

outputs the first 10 observations, which belong to the setosa species.

1.1

1.1.1

Categorical arrays

Nominal arrays

We will use the species cell array from the fisheriris data to construct a nominal array.
Nominal arrays are constructed with the nominal function. Type:
>> help nominal
to obtain detailed help on the function (see also section 4.1 in the Annex). We will use
here the simplest constructor:
>> ndata = nominal( species );
>> getlabels( ndata )
ans =
setosa versicolor virginica
The labels are assigned in the same sequence as they were found in the species cell
array, but are not given any numerical or lexicographic order. One can override the
name of the labels and their matching to the entries in the species array by specifying
two additional input parameters for the nominal constructor:
>> ndata = nominal( species, ...
{ species1, species2, species3 }, ...
{setosa, versicolor, virginica } );
>> getlabels( ndata )
ans =
species1 species2 species3
The second parameter defines the names that the nominal array will use to categorize
the data. They do not need to match the ones from the data. In our case, species1 is
an alias for setosa, species2 is an alias for versicolor, and species3 is an alias for
virginica. If the second input is set as { }, the label names will be inherited from the
species cell array. The third argument allows to define a mapping from the labels in
the cell array and those in the nominal array. Imagine that two names are swapped in
the species array: you could fix the problem with following call:
>> ndata = nominal( species, ...
{ setosa, versicolor, virginica }, ...
{ versicolor, setosa, virginica } );
You can get a complete list of methods (functions) applicable to a nominal object by
typing:
>> methods( ndata )
Methods for class nominal:
addlevels
display
isundefined
nominal
uint64
cat
ipermute
isvector
subsref
uint8
...

flipud
setdiff
double
numel
cellstr

intersect
subsasgn
getlabels
setlabels
droplevels

DATA ORGANIZATION

and the corresponding help by typing:

help nominal/functionname
1.1.2

Ordinal arrays

An ordinal array is constructed exactly as a nominal array, with the only exception
that the created levels are ordered.
>> odata = ordinal( species );
Type:
>> help ordinal
to obtain detailed help on the function (see also section 4.2 in the Annex). The optional
third input parameter of the ordinal constructor overrides the default order of the labels
(as found in the input data). If we wanted to give an alternative order to the labels, we
could create an ordinal array as follows:
>> odata = ordinal( species, { }, ...
{ virginica, setosa, versicolor } );
>> getlabels( odata )
ans =
virginica setosa versicolor
Here, odata encodes an ordering of levels with virginica < setosa < versicolor.
ordinal data can indeed be sorted by the order of their labels:
>> sortedOdata = sort( odata );
Exercise

while nominal data cannot. Exercise: try.

Since the fisher Iris data is not the best example for discussing ordinal data arrays,
lets move to a better example.
Example
Imagine we have the following scores in an examination (maximum is 30).
>> scores = ...
[ 25 27 4 27 20 7 12 18 28 28 9 28 28 17 24 8 15 27 24 28 ];
One cool feature of the categorical data arrays is the possibility to assign levels and
labels automatically from the data as follows.

>> edges = 0:10:30; % We set scores between 0 and 10 to be poor,

% scores between 10 and 20 to be satisfactory,
% and scores between 20 and 30 to be outstanding
>> labels = { poor, satisfactory, outstanding };
>> result = ordinal( scores, labels, [ ] , edges )

1.2

Dataset arrays

result =
Columns 1 through 6
outstanding outstanding poor outstanding outstanding poor
Columns 7 through 12
satisfactory satisfactory outstanding outstanding poor outstanding
Columns 13 through 18
outstanding satisfactory outstanding poor satisfactory outstanding
Columns 19 through 20
outstanding outstanding
If we want to extract the outstanding scores, we can do it easily:
>> scores( result == outstanding )
ans =
25 27 27 20 28 28 28 28 24 27 24 28
Imagine now that you want to add another label for scores that are under 5.
>> result( scores < 5 ) = abysmal;
Warning: Categorical level abysmal being added.
Notice how MATLAB automatically added the new level abysmal. Now lets check:
>> getlabels( result )
ans =
poor satisfactory outstanding abysmal
>> scores( result == abysmal )
ans =
4
The newly added label gets the highest order. Since this is not the desired behavior, we
might want to reorder the labels as follows:

>> newLabels = { abysmal, poor, satisfactory, outstanding };

>> result = reorderlevels( result, newLabels );
>> getlabels( result )
ans =
abysmal poor satisfactory outstanding

1.2

Dataset arrays

Dataset arrays can be viewed as tables of values, with rows representing different observations or cases and columns representing different measured variables. In this sense,
dataset arrays are analogous to standard MATLAB numerical arrays. Basic methods for

DATA ORGANIZATION

creating and manipulating dataset arrays parallel the syntax of corresponding methods
for numerical arrays.
While each column of a dataset array must be a variable of a single type, each
row may contain an observation consisting of measurements of different types. In this
sense, dataset arrays lie somewhere between variables that enforce complete homogeneity on the data and those that enforce nothing. Because of the potentially heterogeneous nature of the data, dataset arrays have indexing methods with syntax that
parallels corresponding methods for cell and structure arrays.
In contrast to categorical arrays, a dataset array contains the measurement data along with the categorical data that describes it. Categorical
arrays are indeed accessory objects that are used by the dataset array to
categorize its measurements.
1.2.1

Constructing dataset arrays

Complete help for creating a dataset can be obtained by typing:

>> help dataset

Exercise

(see also section 4.3 in the Annex). The dataset class is quite powerful. We will
discuss only one way to create a dataset array here (and leave the following as an
exercise: create a dataset from the Excel file iris.xls and the text file iris.csv that can
be downloaded from the course website1 ):

dataset({var1,name(s)},{var2,name(s)},...,ObsNames,obsNames};
All names are optional, but help organize the data in the dataset object. You can copy
the following code to a script to simplify the (re-)creation of the dataset:
% Fishers Iris data (1936)
load fisheriris
% Create a nominal array called species from species to
% label the measurements
n = { nominal( species ), species };
%
%
%
m

The variable names associated to the measurement

matrix meas are SL, SW, PL, PW (for Sepal and Petal Length
and Width)
= { meas, SL, SW, PL, PW };

% Create the observation names

NumObs = size( meas, 1 );
NameObs = strcat( { Obs }, num2str( ( 1 : NumObs ), %d ) );
% Create the dataset itself
iris = dataset( n, m, ObsNames, NameObs );
1 http://www.fmi.ch/html/technical_resources/microscopy/homepage2/training_material.html

1.2

Dataset arrays

Run the script to create the dataset and then inspect it:
>> iris
iris =
Obs
Obs
Obs
Obs
Obs
...

1
2
3
4
5

species
setosa
setosa
setosa
setosa
setosa

SL
5.1
4.9
4.7
4.6
5

SW
3.5
3
3.2
3.1
3.6

PL
1.4
1.4
1.3
1.5
1.4

PW
0.2
0.2
0.2
0.2
0.2

The content of iris is displayed in tabular form with all rows and columns labeled by
the names we specified.
The methods associated to a dataset object can be obtained as follows:
>> methods( iris );
Methods for class dataset:
cat
display
set
subsasgn
double
horzcat
subsref
datasetfun
numel
size
export
join
unique

get
vertcat
ndims
end
summary
replacedata

length
dataset
single
isempty
disp
sortrows

Static methods:
empty
Help for the various commands can be obtained as follows:
>> help dataset.functionname
With the get and set modes one can access and modify the properties of the array.
The dataset properties that can be accessed are Description, Units, DimNames,
UserData, ObsNames, VarNames. Exercise: set Fishers iris data (1936) as the
dataset description.
1.2.2

Accessing dataset arrays

Dataset arrays support multiple types of indexing. Like MATLABs numerical matrices, parenthesis () indexing is used to access data subsets. Like MATLABs cell and
structure arrays, dot . indexing is used to access data variables and curly brace {}
indexing is used to access data elements.
Use parenthesis indexing to assign a subset of the data in iris to a new dataset array
iris1:
iris1 = iris(1:5,2:3)

Exercise

DATA ORGANIZATION

iris1 =
Obs1
Obs2
Obs3
Obs4
Obs5

SL SW
5.1 3.5
4.9 3
4.7 3.2
4.6 3.1
5 3.6

Similarly, use parenthesis indexing to assign new data to the first variable in iris1:
iris1(:,1) = dataset([5.2;4.9;4.6;4.6;5])
iris1 =
Obs1
Obs2
Obs3
Obs4
Obs5

SL SW
5.2 3.5
4.9 3
4.6 3.2
4.6 3.1
5 3.6

Variable and observation names can also be used to access data:

SepalObs = iris1({Obs1,Obs3,Obs5},SL)
SepalObs =
SL
Obs1 5.1
Obs3 4.7
Obs5 5
The following code extracts the sepal lengths in iris1 corresponding to sepal widths
greater than 3:
BigSWLengths = iris1.SL(iris1.SW > 3)
BigSWLengths =
5.2000
4.6000
4.6000
5.0000
Dot indexing also allows entire variables to be deleted from a dataset array:
iris1.SL = []
iris1 =
Obs1
Obs2
Obs3
Obs4
Obs5

SW
3.5
3
3.2
3.1
3.6

1.2

Dataset arrays

Curly brace indexing is used to access individual data elements. The following are
equivalent:
iris1{1,1}
ans =
3.5
iris1{Obs1,SW}
ans =
3.5

1.2.3

Computing with dataset arrays

The summary( dataset ) function provides summary statistics for the component variables of a dataset array. Lets use the original Iris dataset as created in section 1.2.1.
>> summary( iris )
species: [150x1 nominal]
setosa
versicolor
50
50

virginica
50

SL: [150x1 double]

min
1st Q
4.3000
5.1000

median
5.8000

3rd Q
6.4000

max
7.9000

SW: [150x1 double]

min
1st Q
2
2.8000

median
3

3rd Q
3.3000

max
4.4000

PL: [150x1 double]

min
1st Q
1
1.6000

median
4.3500

3rd Q
5.1000

max
6.9000

PW: [150x1 double]

min
1st Q
0.1000
0.3000

median
1.3000

3rd Q
1.8000

max
2.5000

Notice how the summaries use all measurements (i.e. 150) for each of the variables,
without discriminating between setosa, versicolor, or virginica measurements.
1.2.4

Grouping data

Grouping variables are utility variables used to indicate which elements in a data set
are to be considered together when computing statistics and creating visualizations.
They may be numeric vectors, string arrays, cell arrays of strings, or categorical arrays.
Logical vectors can be used to indicate membership (or not) in a single group.

10

DATA ORGANIZATION

Grouping variables have the same length as the variables (columns) in a data set.
Observations (rows) i and j are considered to be in the same group if the values of the
corresponding grouping variable are identical at those indices. Grouping variables with
multiple columns are used to specify different groups within multiple variables.
To group the observations by species, the following are all acceptable (and equivalent) grouping variables:
>>
>>
>>
>>

group1
group2
group3
group4

=
=
=
=

species;
grp2idx(species);
char(species);
nominal(species);

%
%
%
%

Cell array of strings

Numeric vector
Character array
Categorical array (see section 1.2.1)

The following is a short list of functions that take groups as input parameter (for the
complete list please see the Statistics Toolbox manual, Chapter 2: Section Grouped
Data): anova1, anovan, boxplot, grp2idx, grpstats, gscatter, kruskalwallis, manova1,
tabulate.
1.2.5

Using grouping variables

We will use groups to play with the Fishers Iris dataset created in section 1.2.1. We
will use the nominal array that we added to the iris dataset (iris.species) as a grouping
variable. While species, as a cell array of strings, is itself a grouping variable, the
categorical array has the advantage that it can be easily manipulated with methods of
the categorical class.
Lets see for instance how many measurements for each species we have:
>> tabulate( iris.species )
Value Count Percent
setosa
50 33.33%
versicolor
50 33.33%
virginica
50 33.33%
Compute some basic statistics for the data (median and interquartile range), by group,
using the grpstats function:
>> [order,number,group_median,group_iqr] = ...
grpstats([iris.SL iris.SW iris.PL iris.PW], ...
iris.species,{gname,numel,@median,@iqr})2
order =
setosa
versicolor
virginica
number =
50
50
50
50
50
50
50
50
50
group_median =
5.0000
3.4000
5.9000
2.8000
2 Either

50
50
50
1.5000
4.3500

0.2000
1.3000

function handles (@median) or function names in character arrays (numel) can be used.

1.2

Dataset arrays

6.5000
group_iqr =
0.4000
0.7000
0.7000

11

3.0000

5.5500

2.0000

0.5000
0.5000
0.4000

0.2000
0.6000
0.8000

0.1000
0.3000
0.5000

The statistics appear in 3-by-4 arrays, corresponding to the 3 groups (setosa, versicolor, virginica) and 4 variables (SL, SW, PL, PW) in the data. The order of
the groups (not encoded in the nominal array group) is indicated by the group names
in order.
To improve the labeling of the data, one can use the grpstats function on the dataset
object directly:
>> stats = grpstats(iris,species,{@median,@iqr})
stats =
setosa
versicolor
virginica

species
setosa
versicolor
virginica

GroupCount
50
50
50

setosa
versicolor
virginica

median_SL
5
5.9
6.5

iqr_SL
0.4
0.7
0.7

setosa
versicolor
virginica

median_SW
3.4
2.8
3

iqr_SW
0.5
0.5
0.4

setosa
versicolor
virginica

median_PL
1.5
4.35
5.55

iqr_PL
0.2
0.6
0.8

setosa
versicolor
virginica

median_PW
0.2
1.3
2

iqr_PW
0.1
0.3
0.5

When you call grpstats with a dataset array as an argument, you invoke the grpstats
method of the dataset class, rather than the grpstats function. The method has a slightly
different syntax than the function, but it returns the same results, with better labeling.
The statistics calculated by the summary function for the sepal length of the species
setosa could be called explicitly as follows:
% Minimun
>> min( iris.SL( iris.species == setosa ) )
ans =

12

STATISTICAL VISUALIZATION

4.3000
% First quartile
>> quantile( iris.SL( iris.species == setosa ), 0.25 )
ans =
4.8000
% Median
median( iris.SL( iris.species == setosa ) )
ans =
5
% Third quartile
quantile( iris.SL( iris.species == setosa ), 0.75 )
ans =
5.2000
% Maximum
max( iris.SL( iris.species == setosa ) )
ans =
5.8000

Statistical visualization

Statistics Toolbox data visualization functions add to the extensive graphics capabilities
already in MATLAB.
Scatter plots are a basic visualization tool for multivariate data. They are used
to identify relationships among variables. Grouped versions of these plots use
different plotting symbols to indicate group membership.
Box plots display a five-number summary of a set of data: the median, the two
ends of the interquartile range (the box), and two extreme values (the whiskers)
above and below the box. Because they show less detail than histograms, box
plots are most useful for side-by-side comparisons of two distributions.
Distribution plots help you identify an appropriate distribution family for your
data.

2.1

Scatter plots

A scatter plot is a simple plot of one variable against another that is helpful for investigating relationship among variables. MATLAB offers the functions scatter and
plotmatrix to produce simple scatter plots and scatter plot matrices, where all variables
are plotted against each other in pairs. The statistics toolbox adds the two functions
gscatter and gplotmatrix that implement support for grouped data.

2.1

Scatter plots

13

4.5
setosa
versicolor

Sepal Width

3.5

2.5

4.5

5.5
Sepal Length

6.5

Figure 2: Scatter plot

The scatter plot in Figure 2, created with the gscatter function, shows the correlation between sepal length and sepal width in two species of Iris (setosa and versicolor).
Use ismember to subset the two species from group:
subset = ismember(species,{setosa,versicolor});3
scattergroup = species(subset);
gscatter(iris.SL(subset),iris.SW(subset),scattergroup)
xlabel(Sepal Length)
ylabel(Sepal Width)
From this plot, one can say that (i) setosa sepals tend to be shorter and wider than
versicolor sepals and (ii) sepal lengths and widths appear to correlate with each other
(the larger the sepal length of a flower, the larger its sepal width tends to be).
If we wanted to plot all possible pairs of variables in the Iris dataset against each
other, we could use the gplotmatrix function like this:
>> gplotmatrix( [ iris.SL iris.SW iris.PL iris.PW ], ...
[ iris.SL iris.SW iris.PL iris.PW ], iris.species );
From figure 3 it is clear that versicolor and virginica flowers are more similar to
each other in terms of sepal and petal witdhs and lenghts that they are to setosa flowers.
Moreover, setosa variables tend to correlate less with each other. The points in the
graphs on the diagonal lie on a straight line since they are plotted against themselves.
If one sets the second matrix of measurements to [], gplotmatrix plots the first matrix
3 Alternatively:

subset = iris.species == setosa | iris.species == versicolor;

14

STATISTICAL VISUALIZATION

8
setosa
versicolor
virginica

7
6
5
4
3
2
6
4
2
2
1
0

82

Figure 3: Scatter plot matrix

against itself, and replaces the plots on the diagonals with histograms of the various
measurements (Figure 4).
>> gplotmatrix( [ iris.SL iris.PL ], [ ], iris.species );

2.2

Box plots

The graph in Figure 5, created with the boxplot command, compares petal lengths in
samples from the three species of iris.
>> boxplot([iris.PL(iris.species==setosa), ...
iris.PL(iris.species==versicolor), ...
iris.PL(iris.species==virginica)], notch, on, ...
labels, {setosa,versicolor,virginica});
This plot has the following features:
The tops and bottoms of each box are the 25th and 75th percentiles of the
samples (also called, as we saw, the first and third quartile), respectively. The
distances between the tops and bottoms are the interquartile ranges.
The line in the middle of each box is the sample median. If the median is not
centered in the box, it shows sample skewness.
The whiskers are lines extending above and below each box. Whiskers are drawn
from the ends of the interquartile ranges to the furthest observations within the
whisker length (the adjacent values).

2.2

Box plots

15

8
setosa
versicolor
virginica

7
6
5

7
6
5
4
3
2
1
5

Figure 4: Scatter plot matrix with histograms

1
setosa

versicolor

Figure 5: Box plot

virginica

16

STATISTICAL VISUALIZATION

Normal Probability Plot

0.99
0.98
0.95
0.90

Probability

0.75

0.50

0.25

0.10
0.05
0.02
0.01

8.5

9.5

10

10.5
Data

11

11.5

12

12.5

Figure 6: Normal probability plot of a normally-distributed sample

Observations beyond the whisker length are marked as outliers. By default, an
outlier is a value that is more than 1.5 times the interquartile range away from
the top or bottom of the box, but this value can be adjusted with additional input
arguments. Outliers are displayed with a red + sign.
Notches display the variability of the median between samples. The width of a
notch is computed so that box plots whose notches do not overlap (as in Figure
5) have different medians at the 5% significance level. The significance level
is based on a normal distribution assumption, but comparisons of medians are
reasonably robust for other distributions. Comparing box-plot medians is like a
visual hypothesis test, analogous to the t-test used for means.

2.3

Distribution plots

The only probability plot we will discuss here is the normal probability plot. Normal probability plots are used to assess whether data comes from a normal distribution. Many statistical procedures make the assumption that an underlying distribution
is normal, so normal probability plots can provide some assurance that the assumption
is justified, or else provide a warning of problems with the assumption.
First we plot data sampled from a normal distribution to see what we should expect.
>> x = normrnd(10,1,25,1);
>> normplot(x);
Exercise

It is left as an exercise to figure how how the normrnd function works (help normrnd).

17

Normal Probability Plot

0.997
0.99
0.98
0.95
0.90

Probability

0.75
0.50
0.25
0.10
0.05
0.02
0.01
0.003
0

10

20

30

40

50

Data

Figure 7: Normal probability plot of an exponentially-distributed sample

The plus signs in Figure 6 plot the empirical probability versus the data value for
each point in the data. A solid line connects the 25th and 75th percentiles in the data,
and a dashed line extends it to the ends of the data. The y-axis values are probabilities
from zero to one, but the scale is not linear. The distance between tick marks on the yaxis matches the distance between the quantiles of a normal distribution. The quantiles
are close together near the median (probability = 0.5) and stretch out symmetrically as
you move away from the median.
In a normal probability plot, if all the data points fall near the line, an assumption
of normality is reasonable. Otherwise, the points will curve away from the line, and an
assumption of normality is not justified.
For example:
>> x = exprnd(10,100,1);
>> normplot(x);
The plot in Figure 7 is strong evidence that the underlying distribution is not normal.
Exercise: plot the sepal lengths for the three iris species in a normal probability
plot and discuss the outcome. Does the normality assumption hold?

References
1. The official Statistics Toolbox documentation:
http://www.mathworks.com/access/helpdesk/help/toolbox/stats/ (html).
http://www.mathworks.com/access/helpdesk/help/pdf_doc/stats/stats.pdf (pdf)

Exercise

18

ANNEX

Annex

The help information for the classes nominal, ordinal and dataset are printed here for
reference.

4.1

help nominal
NOMINAL Create a nominal array.
B = NOMINAL(A) creates a nominal array from A. A is a numeric, logical,
character, or categorical array, or a cell array of strings. NOMINAL
creates levels of B from the sorted unique values in A, and creates
default labels for them.
B = NOMINAL(A,LABELS) creates a nominal array from A, labelling the levels
in B using LABELS. LABELS is a character array or cell array of strings.
NOMINAL assigns the labels to levels in B in order according to the sorted
unique values in A.
B = NOMINAL(A,LABELS,LEVELS) creates a nominal array from A, with possible
levels defined by LEVELS. LEVELS is a vector whose values can be compared
to those in A using the equality operator. NOMINAL assigns labels to each
level from the corresponding elements of LABELS. If A contains any values
not present in LEVELS, the levels of the corresponding elements of B are
undefined. Pass in [] for LABELS to allow NOMINAL to create default labels.
B = NOMINAL(A,LABELS,[],EDGES) creates a nominal array by binning the
numeric array A, with bin edges given by the numeric vector EDGES. The
uppermost bin includes values equal to the rightmost edge. NOMINAL
assigns labels to each level in B from the corresponding elements of
LABELS. EDGES must have one more element than LABELS.
By default, an element of B is undefined if the corresponding element of A
is NaN (when A is numeric), an empty string (when A is character), or
undefined (when A is categorical). NOMINAL treats such elements as
"undefined" or "missing" and does not include entries for them among the
possible levels for B. To create an explicit level for those elements
instead of treating them as undefined, you must use the LEVELS input, and
include NaN, the empty string, or an undefined element.
You may include duplicate labels in LABELS in order to merge multiple
values in A into a single level in B.
See also ordinal, histc.
Reference page in Help browser
doc nominal

4.2

help ordinal
ORDINAL Create an ordinal array.
B = ORDINAL(A) creates an ordinal array from A. A is a numeric, logical,
character, or categorical array, or a cell array of strings. ORDINAL
creates levels of B from the sorted unique values in A, and creates
default labels for them.
B = ORDINAL(A,LABELS) creates an ordinal array from A, labelling the levels
in B using LABELS. LABELS is a character array or cell array of strings.
ORDINAL assigns the labels to levels in B in order according to the sorted
unique values in A.
B = ORDINAL(A,LABELS,LEVELS) creates an ordinal array from A, with
possible levels and their order defined by LEVELS. LEVELS is a vector
whose values can be compared to those in A using the equality operator.
ORDINAL assigns labels to each level from the corresponding elements of
LABELS. If A contains any values not present in LEVELS, the levels of the
corresponding elements of B are undefined. Pass in [] for LABELS to allow
ORDINAL to create default labels.
B = ORDINAL(A,LABELS,[],EDGES) creates an ordinal array by binning the

4.3

help dataset

numeric array A, with bin edges given by the numeric vector EDGES. The
uppermost bin includes values equal to the rightmost edge. ORDINAL
assigns labels to each level in B from the corresponding elements of
LABELS. EDGES must have one more element than LABELS.
By default, an element of B is undefined if the corresponding element of A
is NaN (when A is numeric), an empty string (when A is character), or
undefined (when A is categorical). ORDINAL treats such elements as
"undefined" or "missing" and does not include entries for them among the
possible levels for B. To create an explicit level for those elements
instead of treating them as undefined, you must use the LEVELS input, and
include NaN, the empty string, or an undefined element.
You may include duplicate labels in LABELS in order to merge multiple
values in A into a single level in B.
See also nominal, histc.
Reference page in Help browser
doc ordinal

4.3

help dataset
DATASET Create a dataset array.
DS = DATASET(VAR1, VAR2, ...) creates a dataset array DS from the
workspace variables VAR1, VAR2, ... . All variables must have the same
number of rows.
DS = DATASET(..., {VAR,name}, ...) creates a dataset variable named
name in DS. Dataset variable names must be valid MATLAB identifiers,
and unique.
DS = DATASET(..., {VAR,name1,...,name_M}, ...), where VAR is an
N-by-M-by-P-by-... array, creates M dataset variables in DS, each of size
N-by-P-by-..., with names name1, ..., name_M.
DS = DATASET(..., VarNames, {name1, ..., name_M}) creates dataset
variables that have the specified variable names. The names must be valid
MATLAB identifiers, and unique. You may not provide both the VarNames
parameter and names for individual variables.
DS = DATASET(..., ObsNames, {name1, ..., name_N}) creates a dataset
array that has the specified observation names. The names need not be
valid MATLAB identifiers, but must be unique.
Dataset arrays can contain variables that are built-in types, or objects that
are arrays and support standard MATLAB parenthesis indexing of the form
var(i,...), where i is a numeric or logical vector that corresponds to
rows of the variable. In addition, the array must implement a SIZE method
with a DIM argument, and a VERTCAT method.
You can also create a dataset array by reading from a text or spreadsheet
file, as described below. This creates scalar-valued dataset variables,
i.e., one variable corresponding to each column in the file. Variable
names are taken from the first row of the file.
DS = DATASET(File,FILENAME, ...) creates a dataset array by reading
column-oriented data in a tab-delimited text file. The dataset variables
that are created are either double-valued, if the entire column is
numeric, or string-valued, i.e. a cell array of strings, if any element in
a column is not numeric. Fields that are empty are converted to either
NaN (for a numeric variable) or the empty string (for a string-valued
variable). Insignificant whitespace in the file is ignored.
Specify a delimiter character using the Delimiter parameter name/value
pair. The delimiter can be any of , \t, ,, ;, | or their
corresponding string names space, tab, comma, semi, or bar.
Specify strings to be treated as the empty string in a numeric column
using the TreatAsEmpty parameter name/value pair. This may be a
character string, or a cell array of strings. TreatAsEmpty only applies
to numeric columns in the file, and numeric literals such as -99 are not
accepted.

19

20

ANNEX

DS = DATASET(File,FILENAME,Format,FORMAT, ...) creates a dataset

array using the TEXTSCAN function to read column-oriented data in a text
file. FORMAT is a format string as accepted by the TEXTSCAN function.
You may also specify any of the parameter name/value pairs accepted by
the TEXTSCAN function.
DS = DATASET(XLSFile,XLSFILENAME, ...) creates a dataset array from
column-oriented data in an Excel spreadsheet file. You may also specify
the Sheet and Range parameter name/value pairs, with parameter values
as accepted by the XLSREAD function. Variable names are taken from the
first row of the spreadsheet. If the spreadsheet contains figures or
other non-tabular information, you should use the Range parameter to
read only the tabular data. By default, the XLSFile option reads data
from the spreadsheet contiguously out to the right-most column that
contains data, including any empty columns that precede it. If the
spreadsheet contains one or more empty columns between columns of data,
use the Range parameter to specify a rectangular range of cells from
which to read variable names and data.
When reading from a text or spreadsheet file, the ReadVarNames parameter
name/value pair determines whether or not the first row of the file is
treated as variable names. Specify as a logical value (default true).
When reading from a text or spreadsheet file, the ReadObsNames parameter
name/value pair determines whether or not the first column of the file is
treated as observation names. Specify as a logical value (default false).
If the ReadVarNames and ReadObsNames parameter values are both true,
the name in the first column of the first row of the file is saved as the
first dimension name for the dataset.
See also dataset/set, dataset/get, genvarname, tdfread, textscan, xlsread.
Reference page in Help browser
doc dataset