Car Quality Prediction Based on Multiple Machine Learning Algorithms

As technology advances, the role that cars play in our daily lives increases. Both car manufactures and customers are eager to know about the car quality to help them choose the car they like. In this paper, three models were used to train and find the relationship between the car attributes and labels to predict the car quality. The dataset was found in the UCI dataset at first. Next the attributes and label of the data were changed into string type to fulfill the requirement of training the machine learning model. Then the dataset was divided into the training set and testing set to fit the model. The machine learning algorithms used in the article includes the Decision Tree model, Random Forest model, Extra Tree model. These models are used to train and test the dataset respectively and the accuracy of each model can be obtained. Finally, the most suitable hyperparameters in each model were found by using the validation curve. Experimental results indicate that the Random Forest model’s precision can reach to 85.66% and it is the most accurate one among three models. Thus, The Random Forest model would help the customers and manufacturers to have the better car.