This is the PR to add my project Car price prediction

car-price-prediction/README.md

@@ -0,0 +1,56 @@
+ # CarCast - Used Car Price Prediction
+
+CarCast is an end-to-end data science + web application project that estimates the fair market price of used cars in India using machine learning.
+
+This project starts from real listing data (sourced from CarDekho), goes through data understanding, cleaning, preprocessing, model training, feature engineering, model re-training, and finally serves predictions through a FastAPI backend with a React frontend.
+
+## What this project does
+
+- Predicts used car selling price from core vehicle attributes.
+- Provides deal analysis by comparing predicted value vs asking price.
+- Exposes a clean API for prediction and model metadata.
+- Delivers a user-facing web app where anyone can enter car details and get an instant estimate.
+
+## Data snapshot
+
+- Source dataset: `cardekho_dataset.csv` (CarDekho listings)
+- Total raw records: **15,411**
+- Raw columns: **14**
+- Cleaned dataset: **15,411 x 13**
+- Feature-engineered dataset: **15,411 x 22**
+- Train/Test split (enhanced processed): **12,328 / 3,083**
+
+## ML journey in short
+
+1. Collected and explored used-car listing data.
+2. Cleaned and preprocessed data.
+3. Trained multiple baseline regressors.
+4. Performed feature engineering to better capture depreciation, usage intensity, and brand/engine interactions.
+5. Repeated preprocessing and model selection on enhanced data.
+6. Selected **XGBoost Regressor** as final model and saved the full prediction pipeline.
+
+## Final selected model
+
+**Model:** XGBoost Regressor  
+**R2:** 0.9459  
+**MAE:** 97,149.24  
+**RMSE:** 201,768.88
+
+The final model performance report is available in [model_performance_enhanced.csv](D:/Car-price-prediction-project.worktrees/car-price-prediction-documentation/backend/reports/model_performance_enhanced.csv).
+
+## Tech stack
+
+- **Data Science:** Python, Pandas, NumPy, scikit-learn, XGBoost, Matplotlib, Seaborn, Jupyter
+- **Backend/API:** FastAPI, Pydantic, Uvicorn, Joblib
+- **Frontend:** React, TypeScript, TanStack Router, TanStack Query, Tailwind CSS
+
+## Important project resources
+
+- Data files: [backend/data/](D:/Car-price-prediction-project.worktrees/car-price-prediction-documentation/backend/data)
+- Notebooks (full DS workflow): [backend/notebooks/](D:/Car-price-prediction-project.worktrees/car-price-prediction-documentation/backend/notebooks)
+- Saved model pipelines: [backend/models/](D:/Car-price-prediction-project.worktrees/car-price-prediction-documentation/backend/models)
+- Full DS write-up: [documentation.md](D:/Car-price-prediction-project.worktrees/car-price-prediction-documentation/documentation.md)
+
+## Scope note
+
+This is a strong portfolio data science project focused on model development and practical deployment. It is not positioned as a production-grade automotive pricing platform.

car-price-prediction/backend/README.md

@@ -0,0 +1,32 @@
+# Backend (FastAPI + ML Inference)
+
+This backend serves the trained used-car price prediction pipeline and exposes APIs for prediction, deal analysis, health checks, and model metadata.
+
+## Quick setup
+
+```bash
+cd backend
+python -m venv .venv
+.venv\Scripts\activate
+pip install -r requirements.txt
+uvicorn app.main:app --host 0.0.0.0 --port 8000 --reload
+```
+
+API docs:
+- Swagger: `http://localhost:8000/docs`
+- ReDoc: `http://localhost:8000/redoc`
+
+## Main endpoints
+
+- `GET /health`
+- `GET /brands`
+- `GET /brands/{brand}/models`
+- `GET /model-metrics`
+- `GET /model-info`
+- `POST /predict`
+- `POST /predict-with-asking`
+
+## Notes
+
+- Final model artifact is loaded from [full_pipeline.pkl](D:/Car-price-prediction-project.worktrees/car-price-prediction-documentation/backend/models/full_pipeline.pkl).
+- Data science notebooks and dataset files are in [backend/notebooks/](D:/Car-price-prediction-project.worktrees/car-price-prediction-documentation/backend/notebooks) and [backend/data/](D:/Car-price-prediction-project.worktrees/car-price-prediction-documentation/backend/data).

car-price-prediction/backend/app/config.py

@@ -0,0 +1,56 @@
+import os
+from pathlib import Path
+from typing import List, Dict
+
+# Base directory
+BASE_DIR = Path(__file__).resolve().parent.parent
+
+# Model paths
+MODEL_PATH = BASE_DIR / "models" / "full_pipeline.pkl"
+
+# Data paths
+DATA_DIR = BASE_DIR / "app" / "data"
+BRANDS_PATH = DATA_DIR / "brands.json"
+BMODELS_PATH = DATA_DIR / "models.json"
+FEATURE_IMPORTANCE_PATH = DATA_DIR / "feature_importance.json"
+
+# Luxury brands list
+LUXURY_BRANDS = [
+    'Mercedes-Benz', 'BMW', 'Audi', 'Land Rover',
+    'Jaguar', 'Porsche', 'Lexus', 'Volvo', 'Mini',
+    'Mercedes-AMG', 'Maserati', 'Bentley', 'Rolls-Royce',
+    'Ferrari', 'Lamborghini', 'Mclaren', 'Aston Martin'
+]
+
+# Engine category bins
+ENGINE_BINS = [0, 1500, 2500, 3500, 6000]
+ENGINE_LABELS = ['Small', 'Mid', 'Large', 'Luxury']
+
+# API settings
+API_TITLE = "CarCast - Used Car Price Prediction API"
+API_VERSION = "2.0.0"
+API_DESCRIPTION = """
+## CarCast API
+
+A comprehensive API for used car price prediction with advanced features.
+
+### Features:
+- **Car Price Prediction**: Predict fair market price
+- **Deal Analysis**: Compare predicted vs asking price
+- **Model Metrics**: View model performance
+- **Feature Importance**: Understand what drives prices
+- **Brand/Model Data**: Get available brands and models
+- **Prediction History**: Track past predictions
+"""
+
+# Model performance metrics (from your training)
+MODEL_METRICS = {
+    "model_name": "XGBoost Regressor",
+    "r2_score": 0.9459,
+    "mae": 97149.24,
+    "rmse": 201768.88,
+    "training_samples": 12328,
+    "test_samples": 3083,
+    "features_count": 295,
+    "description": "XGBoost model trained on 15,240 used car listings with 11 base features and 9 engineered features"
+}

car-price-prediction/backend/app/data/brands.json

@@ -0,0 +1,34 @@
+[
+    "Audi",
+    "BMW",
+    "Bentley",
+    "Datsun",
+    "Ferrari",
+    "Force",
+    "Ford",
+    "Honda",
+    "Hyundai",
+    "ISUZU",
+    "Isuzu",
+    "Jaguar",
+    "Jeep",
+    "Kia",
+    "Land Rover",
+    "Lexus",
+    "MG",
+    "Mahindra",
+    "Maruti",
+    "Maserati",
+    "Mercedes-AMG",
+    "Mercedes-Benz",
+    "Mini",
+    "Nissan",
+    "Porsche",
+    "Renault",
+    "Rolls-Royce",
+    "Skoda",
+    "Tata",
+    "Toyota",
+    "Volkswagen",
+    "Volvo"
+]