Prediction Of Used Car Prices Using Artificial Neural Networks And Machine Learning

With the extensive growth in usage of cars, the newly produced cars are unable to reach the customers for various reasons like high prices, less availability, financial incapability, and so on. Hence the used car market is escalated across the globe but in India, the used car market is in a very nascent stage and mostly dominated by the unorganized sector. This gives chance for fraud while buying a used car. Hence a high precision model is required which will estimate the price of an used car with none bias towards customer or merchandiser.In this model, A Supervised learning-based Artificial Neural Network model and Random Forest Machine Learning model are developed which can learn from the car dataset provided to it. This project presents a working model for used car price prediction with a low error value. A considerable number of distinct attributes are examined for reliable and accurate predictions. The results obtained agree with theoretical predictions and have shown improvement over models which use simple linear models. An ANN (Artificial Neural Network) is built by using Keras Regression algorithm namely Keras Regressor and other Machine Learning Algorithms namely Random Forest, Lasso, Ridge, Linear regressions are built. These algorithms are tested with the car dataset. Experimental results have shown that the Random Forest model with a Mean Absolute Error value of 1.0970472 and R2 error value of 0.772584 has given the less error among all the other algorithms. The work presented here has shown profound implications for future studies of Used Cars price Prediction using Random Forest and might one day help to solve the problem of fraudswith one hundred percent accuracy.