SUML

SUML - Car Price Prediction

Machine Learning course project at PJATK. Predicts used car prices using AutoGluon ensemble models.

Setup

Clone from Hugging Face

This repository is hosted on Hugging Face Hub to handle large model and data files.

# Install git-lfs (required for large files)
# Ubuntu/Debian: sudo apt install git-lfs
# Fedora: sudo dnf install git-lfs
# macOS: brew install git-lfs
# Windows: Download from https://git-lfs.com

git lfs install
git clone https://huggingface.co/bunny501/SUML
cd SUML

Alternatively, use the Hugging Face CLI:

pip install huggingface_hub
huggingface-cli download bunny501/SUML --local-dir ./SUML
cd SUML

Install Dependencies

python -m venv .venv

# Linux/macOS:
source .venv/bin/activate

# Windows:
.venv\Scripts\activate

pip install -r requirements.txt

Running the App

streamlit run App/main.py

Project Structure

SUML/
├── App/                          # Streamlit web application
│   ├── main.py                   # Main UI - form inputs and prediction display
│   ├── inference.py              # Model loading and prediction logic
│   ├── feature_defaults.json     # Default feature values (dataset averages)
│   ├── make_model_mapping.json   # Car make/model dropdown data
│   └── column_value_ranges.json  # Valid ranges for input validation
│
├── AutogluonModels/              # Trained model files (WeightedEnsemble_L2)
│
├── Data/                         # Datasets
│   ├── Cleaned_train.csv         # Main training dataset
│   ├── sales_ads_train.csv       # Raw training data
│   ├── sales_ads_test.csv        # Raw test data
│   └── synthetic_*.csv           # Synthetic data (MostlyAI, SDV)
│
├── src/                          # Source modules
│   └── Autogluon.py              # Model training configuration and utilities
│
├── Notebooks/                    # Jupyter notebooks
│   ├── EDA.ipynb                 # Exploratory data analysis
│   └── ValueRangeExtraction.ipynb
│
├── Predicting-and-Analyzing.../  # Reference project with XGBoost experiments
│   ├── DataCleaning.ipynb
│   ├── DataExploration.ipynb
│   ├── Prediction.ipynb
│   └── ...
│
├── DatasetCleanUpPreparation.py  # Data preprocessing script
├── requirements.txt              # Python dependencies
└── README.md

Training a New Model

source .venv/bin/activate
python -c "from src.Autogluon import run_exploration; run_exploration()"

Features

The model uses:

• Basic info: year, mileage, condition, body type, doors
• Engine specs: fuel type, power (HP), engine size, transmission, drivetrain
• 20 equipment features: leather, heated seats, AC, cruise control, alloy wheels, LED lights, parking sensors, GPS, Bluetooth, etc.
• Standard features (auto-assumed): ABS, airbags, central locking, power steering