Car Price Prediction

A complete end-to-end machine learning pipeline for predicting the resale value of used cars. From data exploration and preprocessing, through model training and evaluation, to serving predictions via a REST API and containerized deployment.

---

🚀 Project Features

• Exploratory Data Analysis (EDA)
  Investigate data integrity, feature distributions, correlations, and outliers to guide modeling decisions.

• Data Preprocessing & Feature Engineering
  Automatic handling of missing values, categorical encoding, numerical scaling, and derivation of new features such as car age.

• Model Development & Hyperparameter Tuning
  Baseline linear regression plus tree-based regressors (Random Forest, Gradient Boosting) with cross-validated grid search.

• Model Evaluation
  Performance metrics (MAE, RMSE, R²), residual analysis, and feature-importance visualizations.

• Model Serialization
  A single, versioned sklearn.Pipeline (including preprocessing) saved via joblib for reproducible inference.

• REST API
  Lightweight FastAPI/Flask service exposing a /predict endpoint for real-time price estimates.

• Docker Support
  Dockerfile for containerized deployment on platforms such as Heroku, AWS ECS, or Azure Web Apps.

---

📁 Repository Structure

/ ├── data/ # Raw and processed datasets ├── images/ # Static assets (figures, banners) ├── notebooks/ # Exploratory Data Analysis (EDA) notebooks ├── src/ # Application code │ ├── data_prep.py # Data cleaning & feature‐engineering │ ├── train.py # Model training & hyperparameter tuning │ ├── evaluate.py # Evaluation metrics & plots │ └── app.py # Flask web application ├── requirements.txt # Python dependencies └── README.md # This file

🔍 Overview

We train regression models (Linear Regression, Random Forest, Gradient Boosting) on a used-car dataset to learn how features like age, mileage, engine size, make/model and fuel type influence resale price. After evaluating performance (MAE, RMSE, R²), the best model is serialized and served via a lightweight Flask API.

---

⚙️ Setup & Installation

1. Clone the repository and navigate into the project folder.
2. Create a Python virtual environment and activate it.
3. Install all required packages from requirements.txt.

---

📊 Exploratory Data Analysis

Open the eda.ipynb notebook under notebooks/.

• Perform data quality checks (missing values, duplicates, outliers).
• Visualize distributions for numeric and categorical features.
• Generate correlation matrices and scatter plots (e.g. price vs. mileage, price vs. age) to inform feature engineering.

---

🏋️ Model Training & Evaluation

1. Prepare and clean your processed dataset (under data/processed/).
2. Run the training script in src/train.py to fit your regression pipelines and perform hyperparameter tuning.
3. Run the evaluation script in src/evaluate.py to compute MAE, RMSE and R² on hold-out data and save residual plots.

Outputs:

• models/car_price_pipeline.joblib — serialized sklearn Pipeline
• reports/metrics.json — performance metrics
• Residual plots in the reports/ folder

---