# Residential Property Value Prediction

## Overview

This competition challenges participants to predict residential property values based on a comprehensive dataset of home characteristics. The competition is part of two Washington University in St. Louis courses: [T81-558: Applications of Deep Neural Networks](https://github.com/jeffheaton/app_deep_learning) and [T81-559: Applications of Generative AI](https://github.com/jeffheaton/app_generative_ai), taught by instructor [Jeff Heaton](https://sites.wustl.edu/jeffheaton/).

## The Challenge

Participants will predict a property value score (0-100) that represents the relative value of residential properties. This score is influenced by various factors including:

* Major factors: environment type, total area, construction year, energy efficiency, smart home features, and room counts
* Minor factors: architectural style, materials quality, landscaping, and garage capacity

## Dataset Description

This unique dataset combines structured, unstructured, and image data:

### Structured Data
* Physical characteristics (area, bedroom count, bathroom count)
* Construction details (year built, materials, energy rating)
* Property features (garage capacity, smart home level)

### Unstructured Data
* Detailed textual descriptions of interior and exterior features
* Architectural style descriptions
* Landscape and environment descriptions

### Image Data
* Exterior photographs of properties (Note: not all properties may have associated images)

![Example House](https://s3.us-east-1.amazonaws.com/data.heatonresearch.com/kaggle/images/2025_fall/10.jpg)

## Approach Opportunities

The competition offers multiple paths to success:

* Deep Learning students may focus on combining structured data with computer vision approaches using the property images
* Generative AI students may emphasize natural language processing of the detailed property descriptions
* The most competitive solutions will likely integrate all three data types: structured, unstructured, and image data

## Target Variable

The target variable is a property value score ranging from 0 to 100, where higher scores indicate more valuable properties. 

While this competition is primarily for students in the specified Washington University courses, it remains open to the broader Kaggle community.