Car Trading Cycle Prediction based on Random Forest Algorithm

This paper uses machine learning to predict the second-hand car trading cycle. In order to enhance the expressivity of the model, we processed the original data by merging data, dividing data into boxes, creating new features, processing outliers, and using principal component analysis to reduce the dimension of features. Then, a random forest was used for model training, and the prediction data of the vehicle transaction cycle was fitted by the 50% discount cross-validation method. The relationship between each feature and the second-hand car transaction cycle was obtained. Finally, the prediction model of the second-hand car transaction cycle is established. In this paper, the mean absolute error (MAE) of the fusion model is 4.72 in the training set and 10.32 in the test set. The model has achieved high accuracy in predicting the vehicle transaction cycle.