A Classification and Price Prediction of Used Cars using Supervised Machine Learning

Demand for used-car price prediction research has grown, which is positive for the business. People in poor nations prefer secondhand automobiles, thus there are few new ones. Important vehicle data is generated by suppliers. Every seller requires a dependable price prediction method to maximize profits. Data-driven algorithms consider the vehicle's production year and mileage to price used cars in India. This was achieved by mining data from the online shop. Data processing reduced duplicates and gaps. We utilize CarDekho.com data to estimate used car pricing components to construct strong Machine Learning (ML) models for using K-Nearest Neighbor (KNN) and Naïve Bayes (NB). The model performed year of production, miles, gasoline type, seller type, gearbox type, prior owners, mileage, engine size, and maximum power are listed. We translated units to numbers and calculated the car's age during data processing. An exploratory study shows that age, brand, and power determine automobile costs more than seat capacity and engine size. The researchers examined transformational and variable selection models. The ML model explained 90% of used vehicle price predictions. After validation, the model showed a 10% MAPE between real and anticipated prices. Used vehicle buyers and sellers may benefit from KNN and NB price forecasts. Data-driven methods are needed to optimize pricing and navigate market prediction. The performance of K-Nearest Neighbors (KNN) and Naïve Bayes (NB) algorithms using three error metrics: Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE). KNN performs better than NB, having lower error values across all three metrics