Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 35 views

    Assignment 2

    Uploaded by

    T M
    This document describes an assignment to implement and compare linear regression and ridge regression models to predict purchasing power. The assignment involves exploratory data analysis, implementing closed-form, gradient descent, and ridge regression models from scratch and with scikit-learn, experimenting with hyperparameters, and reporting results on test data. The goal is to help a trading company predict customer purchasing power in different cities to inform investment decisions.

    Copyright:

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

    Assignment 2

    Uploaded by

    T M
    35 views3 pages
    This document describes an assignment to implement and compare linear regression and ridge regression models to predict purchasing power. The assignment involves exploratory data analysis, implementing closed-form, gradient descent, and ridge regression models from scratch and with scikit-learn, experimenting with hyperparameters, and reporting results on test data. The goal is to help a trading company predict customer purchasing power in different cities to inform investment decisions.

    Original Description:

    The best assignent on Machine Learning Foundation and Applications

    Original Title

    Assignment-2

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document describes an assignment to implement and compare linear regression and ridge regression models to predict purchasing power. The assignment involves exploratory data analysis, implementing closed-form, gradient descent, and ridge regression models from scratch and with scikit-learn, experimenting with hyperparameters, and reporting results on test data. The goal is to help a trading company predict customer purchasing power in different cities to inform investment decisions.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    35 views3 pages

    Assignment 2

    Uploaded by

    T M
    This document describes an assignment to implement and compare linear regression and ridge regression models to predict purchasing power. The assignment involves exploratory data analysis, implementing closed-form, gradient descent, and ridge regression models from scratch and with scikit-learn, experimenting with hyperparameters, and reporting results on test data. The goal is to help a trading company predict customer purchasing power in different cities to inform investment decisions.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 3

    Assignment-2: Linear and Ridge Regression

    Problem Statement:

    A trading company X needs to predict the purchasing power of consumers in a given city, so as to
    decide the company’s investment in that city. The company has some historical data that has different
    attributes of various cities and corresponding purchasing power of consumers (more details in
    Dataset section). Develop linear and ridge regression methods to help the company in predicting
    purchasing powers of the customers.

    Implementation: [2+4+2+2=10]
    ● Exploratory data analysis and Feature scaling
    ● Implementation of closed form solution approach towards linear regression
    [LIN_MODEL_CLOSED]
    ● Implementation of gradient descent approach towards linear regression [LIN_MODEL_GRAD]
    with Minibatch
    ● Implementation of gradient descent approach towards linear regression with regularization
    (ridge regression) with Minibatch [LIN_MODEL_RIDGE]

    **Implement [LIN_MODEL_CLOSED] from scratch. You may make use of the numpy library to
    perform matrix operations.

    ** For implementation of [LIN_MODEL_GRAD] and [LIN_MODEL_RIDGE], you may use the


    scikit-learn library.

    **In general, you may use libraries to process and handle data.

    **Experiment 3, 4 and 5 should be done with feature scaled data.

    **For training [LIN_MODEL_GRAD] and [LIN_MODEL_RIDGE], use minibatch size of 256 and total
    no of epochs 50.

    **Performance Metric to be used for evaluating the models is Mean Square Error (MSE)

    Experiments: [2+2+3+3+3=13]
    The dataset will be split into Train:Validation:Test with 60:20:20 ratio.
    1. Experiment 1: EDA: Show the distribution of the features and their pair-wise correlation

    2. Experiment 2: Report and compare performance (MSE) of [LIN_MODEL_CLOSED] with and


    without feature scaling. Performance should be computed with 20% held out test data.

    3. Experiment 3: Report the effect of varying learning rate in [LIN_MODEL_GRAD] on validation


    data. Choose learning rate values from [1e-5, 1e-4, 1e-3, 1e-2, 0.05, 0.1]. Plot MSE vs
    learning rate. Find the best value of the hyperparameter learning rate.
    4. Experiment 4: Report the effect of varying ridge regression hyperparameter (alpha) in
    [LIN_MODEL_RIDGE] on validation data. Use the learning rate found in Experiment 1.
    Choose alpha values from the range 0.0 to 1.0 with an increment of 0.1. Plot MSE vs alpha.
    Find the best value of the hyperparameter alpha.

    5. Experiment 5:
    a. Derive performance of the three models [LIN_MODEL_CLOSED],
    [LIN_MODEL_GRAD] and [LIN_MODEL_RIDGE] with the optimal hyperparameters
    found in the earlier experiments. Use held out 20% data (test) for estimating the
    performance measured with MSE.
    b. Report your observations with appropriate explanations.

    Datasets:
    This dataset comprises sales transactions captured at a retail store.You can find the dataset here.

    Data Overview:

    ● User_ID: Unique ID of the user


    ● Product_ID: Unique ID of the product.
    ● Gender: indicates the gender of the person making the transaction.
    ● Age: indicates the age group of the person making the transaction.
    ● Occupation: shows the occupation of the user, already labelled with numbers 0 to 20 (e.g., 0 may
    correspond to banker, 1 may correspond to doctor, etc)
    ● City_Category: User's living city category. Cities are categorised into 3 different categories 'A', 'B'
    and 'C'.
    ● Stay_In_Current_City_Years: Indicates how long the user has lived in this city.
    ● Marital_Status: is 0 if the user is not married and 1 otherwise.
    ● Product_Category_1 to _3: Category of the product. All 3 are already labelled with numbers.
    ● Purchase: Purchase amount.

    Handling NaN (Missing) Values:

    1. Features with >50% NaN values can be dropped


    2. For numerical features, the NaN values can be imputed with the following methods:
    a. Mean
    b. Median
    c. Forward fill (fill missing value with the previous data point)
    d. Backward fill (fill missing value with the next data point)
    3. For categorical features:
    a. Mode
    b. Forward fill
    c. Backward fill

    Problem: Predict purchase amount


    Submission:
    A .zip file containing the python source code and a PDF report file. The final name should
    follow the template: <Assign-No>_<Your Roll No>.zip. For example, if your roll no is
    15CE30021, the filename for Assignment 2 will be: Assign-2_15ce30021.zip

    1. A single python code (.py) containing the implementations of the models and experiments with
    comments at function level. The first two lines should contain your name and roll no.

    2. A report [PDF] containing [2 points]


    a. Experiment 1: EDA Plots and correlation heatmap
    b. Experiment 2: Performance values for models with and without feature scaling.
    c. Experiment 3: MSE vs learning rate plot. Also the best choice for the learning rate
    value.
    d. Experiment 4: MSE vs alpha plot. Also the best choice for the alpha value.
    e. Experiment 5: Table capturing comparison among [LIN_MODEL_CLOSED],
    [LIN_MODEL_GRAD] and [LIN_MODEL_RIDGE] followed by your observations

    Responsible TAs:
    Please write to the following TAs for any doubt or clarification regarding Assignment 2
    Aditya Chawla: adityachawla700@gmail.com
    Ramishetti Sai sreeja: ramishetti@iitkgp.ac.in

    Deadline:
    The deadline for submission is 26th August (Saturday), 11:55PM, IST. Irrespective of the time in
    your device, once submission in moodle is closed, no request for submission post-deadline will be
    entertained. No email submission will be considered. So, it is suggested that you start submitting the
    solution at least one hour before the deadline.

    You might also like