Used Cars Price Prediction - README

Overview

This project implements a regression analysis on a dataset of used cars to predict their prices. It uses 15 popular machine learning models and compares their performances using metrics such as R², relative error, and RMSE. Some of the complex models have been optimized for better results.

Table of Contents

1. Features

   • Exploratory Data Analysis (EDA)
     
   • Data Preprocessing and Feature Engineering
     
   • Model Training and Hyperparameter Tuning
     
   • Model Evaluation and Comparison
     
   • Prediction and Insights
     

2. Usage

   • Instructions for running the notebook
     
   • Dependencies
     

3. Models Included

   • Linear Regression
     
   • Support Vector Machines (SVR and Linear SVR)
     
   • Multi-Layer Perceptron Regressor (MLP)
     
   • Stochastic Gradient Descent (SGD)
     
   • Decision Tree and Random Forest Regressors
     
   • XGBoost and LightGBM
     
   • Gradient Boosting Regressor
     
   • Ridge Regressor
     
   • Bagging Regressor
     
   • ExtraTrees Regressor
     
   • AdaBoost Regressor
     
   • Voting Regressor
     

Features

1. Dataset

• The dataset is downloaded and preprocessed to remove unnecessary or redundant columns and handle missing values.
• Target variable: price
  
• Features include year, manufacturer, condition, cylinders, fuel type, odometer reading, transmission type, drive type, vehicle type, and paint color.

2. Preprocessing

• Categorical features are encoded using LabelEncoder.
• Continuous features are scaled using StandardScaler.
• Data is split into training, validation, and testing sets.

3. Model Evaluation

• Models are evaluated using:
  • R² score for goodness of fit
    
  • Relative Error for accuracy
    
  • RMSE for prediction errors
    

Usage

1. Install Dependencies

Ensure the following Python libraries are installed: - Data Handling & Visualization: numpy, pandas, matplotlib - Modeling: sklearn, xgboost, lightgbm - Hyperparameter Tuning: hyperopt

To install the dependencies: bash pip install numpy pandas matplotlib scikit-learn xgboost lightgbm hyperopt

2. Run the Project

The main notebook is main.ipynb. Open it in Jupyter Notebook or JupyterLab, and execute the cells sequentially.

1. Load the Dataset: Ensure the dataset is available in the specified path. Modify the file path in the notebook if required.
2. Explore and Preprocess Data: The notebook performs EDA and data preparation.
3. Train and Evaluate Models: Run each model and observe its performance metrics.
4. Predict Prices: Use the best-performing model for predictions on unseen data.

Results

The best-performing models based on RMSE, R², and relative error are: - LightGBM (LGBM)
- Bagging Regressor
- XGBoost (XGB)

Visualizations

• R² scores, relative errors, and RMSE for all models are plotted for comparison.

Contribution and Feedback

Comments and feedback are welcome to improve the implementation and results.