STAT 159 Group 20 Project: Housing Prices Prediction

Introduction

This is a Final Project of Group 20 for STAT 159 Spring 2023. The project explores factors that impact the housing price, such as demographic, socioeconomic, and environmental factors. The data is obtained from a Kaggle tutorial competition and can be found in the data folder.

A JupyterBook of the main notebook code_notebook.ipynb can be found here.

Project Description

The study starts with a introduction with a brief overview of the project's objectives and the data sources used in the analysis. Then, we import and clean data and create a 70/30 training/test dataset. For exploratory data analysis (EDA), we conduct descriptive statistics and visualizations, such as bar charts, histograms, and scatter plots. With the plots from EDA, we implement feature selection to select the most relevant features for modeling using correlation analysis, feature importance techniques, and domain knowledge.

We also perform a simple ordinary least squares (OLS) regression to predict the outcome of interest using a subset of features. Then, we compare model performance using evaluation metrics such as R-squared, mean squared error, and mean absolute error. We provide the necessary packages and tests for this project.

Installation

Start by tying the following commands in your terminal:

• Create environment
  • Use Makefile: make env
  • Alternatively you can run the conda and python commands yourself
    • To create a new conda environment based on the specifications defined in the environment.yml file:
      mamba env create -f environment.yml -p ~/envs/housing
    • To install a new IPython kernel into the Jupyter Notebook, specifically for the housing environment:
      python -m ipykernel install --user --name housing --display-name "IPython - housing"
    • Activate environment
      conda activate housing
• Install package
  • pip install .
• Build target specified in Makefile
  • make all

Testing

• To run all tests on the functions: pytest tools