Prague Apartment Price Prediction

This repository is dedicated to the development and application of machine learning techniques for predicting the prices of apartments in Prague.

Requirements

---

Ensure you set up a Python environment before running the code. You can use one of the following commands:

1. conda env create -f environment.yml
2. conda env create re && conda activate re && pip install -r requirements.txt

Python Version: The codebase uses Python 3.10.6.

Running the code

---

1. Command Line Interface (CLI)

For detailed instructions, use: python main.py --help

Examples:

Hyperparameter Search: python main.py --train --tune (data loaded from ../data/dataset.csv)

Default Prediction: python main.py (runs prediction on data from ../data/dataset.csv and saves results in ../data/result.csv)

Training with New Data: python main.py --train --scrape (scrapes new data and performs training)

2. Web Interface

Run a local web server: streamlit run web.py

Processing logic

---

The processing runs in two phases:

1. Training Phase: Crawlers obtain all advertisements.
2. Inference Phase: Users provide advertisement URL/data via the web app.

Implementation Details

ETL Class: Handles data acquisition and preprocessing.

Model Class: Operates on data preprocessed by ETL.

The model is based on XGBoost.

Graphical proposal of processing logic

Web interface look

Prediction using manually entered attributes

Prediction using url of sreality advertisement

Effects of attributes on final price prediction

Additional information about neighbourhoods of apartments

Team members

---

• Hanka Nguyenová (Team leader)
• Daniel Karlík
• Emanuel Frátrik
• (Adam Šumník)