June 20, 20252 min read

Car Price Prediction System

Predictive analytics web application leveraging machine learning models for automotive pricing.

Pythonscikit-learnFlaskData AnalysisNginx

Overview

A web application that predicts used car prices based on features like make, model, year, mileage, and condition. The project covers the full pipeline — data collection, cleaning, feature engineering, model training, and production deployment.

Pipeline

Raw Data
    │
    ▼
Data Cleaning (missing values, outliers)
    │
    ▼
Feature Engineering (encoding, scaling)
    │
    ▼
Model Training (Linear Regression, Random Forest, XGBoost)
    │
    ▼
Model Evaluation (RMSE, R²)
    │
    ▼
Export to ONNX / Pickle
    │
    ▼
Flask API + Web UI
    │
    ▼
Nginx → Production

Models Compared

Model	RMSE ($)	R² Score
Linear Regression	3,842	0.82
Random Forest	2,156	0.91
XGBoost	1,893	0.93

XGBoost was chosen for production due to the best accuracy and reasonable inference time.

Deployment

Deployed on the same Linux VPS infrastructure as the Signature Detection System. Containerized with Docker behind Nginx reverse proxy:

# Build and run
docker build -t car-price-predictor .
docker run -d \
  --name car-price-predictor \
  -p 127.0.0.1:5001:5000 \
  --restart unless-stopped \
  car-price-predictor

Nginx config:

server {
    listen 443 ssl http2;
    server_name car.kushaltimilsina.com.np;

    ssl_certificate /etc/letsencrypt/live/car.kushaltimilsina.com.np/fullchain.pem;
    ssl_certificate_key /etc/letsencrypt/live/car.kushaltimilsina.com.np/privkey.pem;

    location / {
        proxy_pass http://127.0.0.1:5001;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header X-Forwarded-Proto $scheme;
    }
}

Key Features

Input vehicle details through a clean web form
Real-time price prediction using the trained XGBoost model
Confidence intervals shown alongside predictions
Responsive design for mobile access