The analysis of the Hungarian used vehicle market is a relevant data mining problem. Customers want to make sound decisions relying on deliberate information when purchasing a used car. Usually they already have an idea about the category, price range and age of the sought vehicle. On the basis of those features the price of the car can be accepted, or refused when a certain car is overestimated. In general the market determines the price of a vehicle precisely. However sometimes also a subjective factor appears in the pricing applied by the vendors, but this can have only slight bias on the actual price.
The price of a used car depends on a couple of attributes. On the basis of the influential attributes models were applied to predict the price. The final aim of the thesis is to create an application, which is able to give precise predictions on the price of used vehicles.
The task was solved in the framework of a complete data mining process based on the CRISP-DM method. After defining the problem and determining the criteria I have downloaded the data from the hasznaltauto.hu. The website covers the whole Hungarian used vehicle market comprising more than 70 thousands of advertisements. The saved data had to be clean, so I applied different cleaning techniques. Besides the cleansing some subpopulations were excluded from the dataset. Models were built using the R language, and they were subsequently evaluated. Finally a user interface was created to the application.
The thesis and the supplementary documents present the above work in details.