Evaluating and comparing prices of used cars can be challenging and very time consuming for customers. I built a project that predicts the price of used cars from 38 different brands based on their make, model, year and mileage. The goal of the project is to estimate the price that a customer should expect to pay for a car based on past selling data. The data for the project was obtained from Kaggle and I used a random forest regressor from Scikit-Learn to build my model. The model was trained with the data of 1.2million used car sells in the US. Finally, I integrated my training weights in an interactive web application. My final product has an average error of 8% between the real price and the predicted price of a car.

Built With

Share this project:
×

Updates