Prediction of The Prices of Second-Hand Cars

In today's economic conditions, interest in second hand products has increased. Especially second-hand car or vehicles have a wide customer base. In the sector which has a workshop market, it is very important to make fast sales, to make the right pricing and to calculate the ideal prices of the cars in order to exchange at the right price. With linear regression analysis second-hand in such cases first determination of variables with effect on price, then it is possible to calculate the price by establising estimating model. In this study, the model was established by determining 23 of 78 variables affecting the price such as price, brands and model years of 5041 second-hand cars. The Determination Rate (R2) of these 23 variables was found to be 89.1%. Then, by using this regression model, second hand prices of the cars were estimated via machine learning algorithm. The data set is divided into two as training and test data (70-30% and 80-20%). As a result of the study, it was determined the affinities between the real values and the estimated values. The proximity rate (±%) calculated in result of study shows affinity intensity of the estimation results to the true results. Via the prediction model established as a result of machine learning, the predictive accuracy rate was found to be 81.15% according to the 10% proximity of the correct results (upper limit; 110%, lower limit; 90%). According to the results, it is thought that machine learning technique could be second-hand to estimate second hand car prices. However, it is possible to reach a better estimation rate with a data set with more units and different variables