Scribd Logo
Upload

Welcome to Scribd!

  • 595 views

    House

    Uploaded by

    Sandy
    This document summarizes an analysis of predicting house sales prices using linear regression and gradient boosting regression models. The analysis used a dataset of over 21,000 home sales in King County, Washington. A linear regression model achieved a prediction accuracy of 73%, while a gradient boosting regression model achieved a higher accuracy of 91.94%, meeting the goal of predicting prices with over 85% accuracy. In conclusion, predicting house prices accurately requires using a large number of house features in a powerful and non-linear model.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd

    House

    Uploaded by

    Sandy
    595 views19 pages
    This document summarizes an analysis of predicting house sales prices using linear regression and gradient boosting regression models. The analysis used a dataset of over 21,000 home sales in King County, Washington. A linear regression model achieved a prediction accuracy of 73%, while a gradient boosting regression model achieved a higher accuracy of 91.94%, meeting the goal of predicting prices with over 85% accuracy. In conclusion, predicting house prices accurately requires using a large number of house features in a powerful and non-linear model.

    Original Description:

    describes the brief description of the project...

    Original Title

    House ppt

    Copyright

    © © All Rights Reserved

    Available Formats

    PPTX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document summarizes an analysis of predicting house sales prices using linear regression and gradient boosting regression models. The analysis used a dataset of over 21,000 home sales in King County, Washington. A linear regression model achieved a prediction accuracy of 73%, while a gradient boosting regression model achieved a higher accuracy of 91.94%, meeting the goal of predicting prices with over 85% accuracy. In conclusion, predicting house prices accurately requires using a large number of house features in a powerful and non-linear model.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    595 views19 pages

    House

    Uploaded by

    Sandy
    This document summarizes an analysis of predicting house sales prices using linear regression and gradient boosting regression models. The analysis used a dataset of over 21,000 home sales in King County, Washington. A linear regression model achieved a prediction accuracy of 73%, while a gradient boosting regression model achieved a higher accuracy of 91.94%, meeting the goal of predicting prices with over 85% accuracy. In conclusion, predicting house prices accurately requires using a large number of house features in a powerful and non-linear model.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    of 19
    At a glance
    Powered by AI
    The presentation discusses predicting house sale prices using linear regression and gradient boosting regression models on a dataset of home sales in King County, Washington.

    The dataset contains features and sale prices of over 21,000 homes sold in King County, Washington between May 2014 and May 2015.

    Linear regression and gradient boosting regression models were used to predict house sale prices based on features like number of bedrooms, square footage, lot size, etc.

    HOUSE SALES PRICE

    PREDICTION ANALYSIS

    Presented by,
    SANDHYA NAIDU E
    • Abstract
    • Project Objective
    • Introduction

    CONTENTS • About Dataset


    • Linear Regression
    • Gradient Booster Regression
    • Snapshots
    • Conclusion
    ABSTRACT

    • This project focuses on predicting the selling price of the house depending on various
    parameters like Year built, Square feet, Lot size, number of beds and baths, features,
    Walk score etc.
    • The data is taken from Kaggle.com
    What is Kaggle ?
    KAGGLE is an online community of data scientist and machine learners owned
    by Google. Kaggle allows users to find and publish data sets, explore and build models in
    a web-based data-science environment, work with other data scientists and machine
    learning engineers, and enter competitions to solve data science challenges.
    PROJECT OBJECTIVE
    • This project aims in constructing a mathematical model using Linear
    Regression and Gradient Booster Regression to estimate the selling price of
    the house based on a set of variables.
    • Analysis Software Used – Python Jupyter Notebook (ANACONDA
    ENVIRNONMENT)
    INTRODUCTION
    • Our goal for this project was to use regression techniques in order to
    estimate the sale price of a house in King County, Washington given the
    feature and pricing data for around 21,000 houses sold within one year.
    ABOUT DATASET
     Our dataset comes from a Kaggle competition.
     The dataset contains house sale prices and its
    features for homes sold in King County,
    Washington between May 2014 and May 2015.
     King County is the most populous county in Washington
    and is included in the Seattle-Tacoma-Bellevue
    metropolitan statistical area. The county is considered
    the 13th most populous county in the United States.
     There are 21,613 observations in the dataset.
     There are 25 total attributes in the dataset, four of which
    we derived from current columns. We are using 22
    attributes in our models: all attributes except for date,
    latitude and longitude
    DATA PREPROCESSING
    Attribute Percentage
    Bedrooms 0.308349598
    Bathroom 0.525137505
    • During cleaning and preprocessing, we created four Sqft_living 0.702035055
    Sqft_lot 0.089660861
    attributes derived from other attributes: Age,
    Floors 0.256793888
    Age_renovated, Sqrt_living15_diff, and Sqft_lot15_diff. Waterfront 0.266369434

    • We chose not to use the variable data, because it only View 0.397293488
    Condition 0.036361789
    shows us when the data was entered into the database. Grade 0.667434256

    • We chose not to use latitude and longitude because the Sqft_above 0.605567298
    Sqft_basement 0.323816021
    attribute, Zip code, contains the same information and was Yr_built 0.054011531
    Age -0.054011531
    easier to work with in our models.
    • We checked for missing variables and the dataset didn’t Yr_renovated
    Age_renovation
    0.126433793
    -0.105754631
    contain any Sqft_living15 0.585378904
    Sqft_living15_diff 0.405391664
    Sqft_lot15 0.082447153
    Sqft_lot15_diff 0.050590661
    LINEAR REGRESSION
    So what did we do ? Let’s go step by step.
    • We import our dependencies , for linear regression we use sklearn (built in python library) and
    import linear regression from it.
    • We then initialize Linear Regression to a variable reg.
    • Now we know that prices are to be predicted , hence we set labels (output) as price columns and we
    also convert dates to 1’s and 0’s so that it doesn’t influence our data much . We use 0 for houses
    which are new that is built after 2014.
    • We again import another dependency to split our data into train and test.
    • I’ve made my train data as 90% and 10% of the data to be my test data , and randomized the
    splitting of data by using random_state.
    • So now , we have train data , test data and labels for both let us fit our train and test data into linear
    regression model.
    • After fitting our data to the model we can check the score of our data ie , prediction. in this case the
    prediction is 73%
    THE ACCURACY OF THE MODEL IS LOWER
    THAN OUR AIM OF 85. SO HOW DO WE
    ACHIEVE THAT 85% TARGET ?
    GRADIENT BOOSTER REGRESSION

    What is Gradient Boosting ?


    It is a machine learning technique for regression and
    classification problems, which produces a prediction
    model in the form of an ensemble of weak prediction
    models, typically decision trees.
    1. We first import the library from sklearn ( it is the best library for all statistical
    related models)
    2. We create a variable where we define our gradient boosting regressor and set
    parameters to it , here
     n_estimator — The number of boosting stages to perform. We should not set it too
    high which would overfit our model.
     max_depth — The depth of the tree node.
     learning_rate — Rate of learning the data.
     loss — loss function to be optimized. ‘ls’ refers to least squares regression
     minimum sample split — Number of sample to be split for learning the data
    3.We then fit our training data into the gradient boosting model and check for
    accuracy
    4.We got an accuracy of 91.94% which is amazing!
    SNAPSHOTS
    CONCLUSION

    In this project, we attempted to predict the price of houses in King County. We found that the best
    we could achieve was a GRADIENT BOOSTER REGRESSION MODEL of an accuracy of 91%
    which we need to consider to be very accurate. Predicting house prices is an extremely complex and
    challenging problems because houses vary widely and house prices are not only based on the
    physical properties of a house but also on the emotional, social and financial position of the parties
    involved. Our results indicate that in order to provide accurate predictions of house prices, a very
    large number of features must be used and that they most likely need to be combined with a
    powerful, complex and non-linear model.
    THANK YOU

    You might also like