🏠 Lublin House Price Prediction

https://davydantoniuk.github.io/lublin-house-price-predict/

This project aims to predict house prices in Lublin using various machine learning and deep learning models. The dataset includes property listings from Otodom web page.

Results

• After comparing various models, the Random Forest Regressor and CatBoost models showed the best performance and were combined in a stacking model.
• The FastAPI app allows easy interaction with the trained model for real-time house price predictions.

Visualizations

1. Model Results Comparison: A bar plot comparing the performance of all models, showcasing their metrics.

2. Prediction Example:

   • Input Example: The last advertisement for sale on otodom.
   • Predicted Output: The model's prediction.
     

📁 Project Structure

1. Introduction

• Overview of the project goals and objectives.

2. Data Collection & Preprocessing

• Scraped property data and handled issues like:
  • Data type conversions
  • Duplicate listings
  • Missing values with KNN imputation
  • One-hot encoding for categorical features

3. Exploratory Data Analysis (EDA)

• Performed detailed analysis on:
  • Price and area distribution
  • Price per sqm
  • Floor and regional analysis
  • Temporal trends and property characteristics

4. Feature Engineering & Data Preparation

• Transformed key features:
  • Floor and regional-based transformation
  • Outlier detection and handling
  • Train-test-validation split

5. Machine Learning Models

• Compared various ML models:
  • Linear Regression, Elastic Net, Decision Trees, Random Forest, Gradient Boosting, XGBoost, LightGBM, CatBoost
  • Built a stacking model for better performance

6. Deep Learning Models

• Developed deep learning models:
  • Feedforward Neural Network
  • Convolutional Neural Network

7. Model Comparison and Interpretation

• Compared performance metrics of all models
• Analyzed feature importance
• Investigated worst predictions

8. Final Model & Predictions

• Saved the best-performing model
• Made predictions on new data

9. FastAPI App

• Developed a FastAPI-based web application to:
  • Serve the best model
  • Provide real-time predictions

Installation & Usage

Run the FastAPI App

1. Clone the repository:

   git clone https://github.com/davydantoniuk/lublin-house-price-predict.git
   cd lublin-house-price-predict/app

2. Install the required packages:

   pip install -r requirements.txt

3. Run the FastAPI app:

   uvicorn app:app --reload

4. Access the app: Open your browser and navigate to http://127.0.0.1:8000

Build the Docker Image

1. Clone the repository:

   git clone https://github.com/davydantoniuk/lublin-house-price-predict.git
   cd lublin-house-price-predict/app

2. Build the Docker image:

   docker build -t lublin-house-price-predict .

3. Run the Docker container:

   docker run -p 80:80 lublin-house-price-predict .

4. Access the app: Open your browser and navigate to http://localhost:80