Comparison of machine learning methods for used car price forecast in the retail industry

In the retail industry, vehicle price estimation is an important tool for developing accurate pricing strategies. Nowadays, thanks to machine learning techniques, it is possible to make price predictions with high accuracy using available data. The study aims to predict vehicle prices using different machine learning algorithms to provide an accurate and effective method in the decision-making processes of vehicle dealers, customers and other stakeholders in the retail sector. The data set used in the study was obtained from the www.arabam.com website with the Uipath tool, known in Robotic Process Automation technology. Models were created using different machine learning techniques such as Artificial Neural Network, XGBoost, Random Forest and Support Vector Machines to predict the price of vehicles. The created models are visualized. Models were compared and the most appropriate models were determined for the examples considered. In comparing the models, performance criteria commonly used in the literature were preferred. Training and test data sets were created with K-fold cross validation and the results were calculated as mean absolute error (MAE), mean squared error (MSE), root mean square error (RMSE) and root mean square error (RMSE). Evaluated by R-squared. In order for users to use the prediction model correctly, a decision support system was created with the Tkinter program, a graphical user interface library used in the Python programming language. Thus, people who want to buy a second-hand vehicle will be able to estimate the price of the vehicle by entering the vehicle features into the decision support system.