BarWachsman7/austin-housing-regression

Assignment 2 — Austin Housing Price Prediction

Project Overview

This project uses the Austin Housing dataset to predict real-estate prices using both regression and classification models.

The first part of the project focuses on predicting the continuous housing price variable: latestPrice.

The second part reframes the problem as a classification task by dividing prices into three tiers:

• Affordable
• Mid-Range
• Luxury

Presentation Video

Watch on YouTube

Dataset

The dataset used in this project is:

austinHousingData.csv

The notebook expects the dataset file to be located in the same directory as the notebook.

Dataset shape:

• 15,171 rows
• 47 columns

Main Notebook

The main notebook file is:

Assignment_2_Final_Version.ipynb

Files Included

This repository includes:

• Assignment_2_Final_Version.ipynb
• austinHousingData.csv
• regression_model.pkl
• classification_model.pkl
• README.md

Regression Task

The regression models compared in this project include:

• Baseline Linear Regression
• Improved Ridge Regression
• Random Forest Regressor
• Gradient Boosting Regressor

Regression Results

Model	R²	MAE
Baseline Linear Regression	0.2347	$163,697.27
Ridge Regression	0.3536	$135,169.93
Random Forest Regressor	0.2446	$118,981.47
Gradient Boosting Regressor	0.2602	$117,041.55

The preferred operational regression model is Gradient Boosting Regressor.

It was selected because it achieved the lowest Mean Absolute Error (MAE), making it the most useful model for practical dollar-level price prediction.

Classification Task

The classification models compared in this project include:

• Logistic Regression
• Random Forest Classifier
• Gradient Boosting Classifier

Classification Results

Model	Macro F1-Score	Macro ROC-AUC
Logistic Regression	0.72	0.8771
Random Forest Classifier	0.79	0.9275
Gradient Boosting Classifier	0.80	0.9362

The preferred classification model is Gradient Boosting Classifier.

It was selected because it achieved the strongest Macro F1-Score and Macro ROC-AUC.

Model Files

Two model files are included:

regression_model.pkl

Contains the final regression model bundle.

classification_model.pkl

Contains the final classification model bundle.

Important note:

The saved pickle models expect the preprocessed feature matrices generated by the notebook, not raw housing data directly.

How to Run

To reproduce the project:

1. Open Assignment_2_Final_Version.ipynb.
2. Make sure austinHousingData.csv is in the same folder.
3. Run all cells from top to bottom.
4. The notebook will train the models and export the pickle files.

Author

Bar Wachsman