Due to the increase in the price of new cars and the lack of funds among users, the sales/purchases of used cars are increasing overall. Predicting the price of used cars attracts a lot of interest in the field of research because it requires considerable effort and knowledge from the expert in the field where the main objective is to use all data generated by vehicle sellers. For the same purpose, specific characteristics are taken into account for precise predictions. The suggested pre-treatment model utilizes a dataset that includes information on the make and model of the vehicle, the year it was produced, its miles, its condition, and other elements that affect used-car prices. To estimate the cost of used cars, the Light Gradient Boosted Machine, Extreme Gradient Boosting, Gradient Boosting Regressor, Random Forest, and Bagging Regressor models are used. Each model was trained using market information gathered from websites. The used techniques to improve predictions are processing techniques on data set and processing regulations by models. We present two different types of data-driven data processing. As a result, we have a new data frame to analyze. We evaluate and compared the data frame with each other of the machine learning models to find the model that works most effectively and consistently. Putting time and effort into preparing the dataset and optimizing the regularization resulted in a decent score with Gradient Boosting Regressor.