Research On Influencing Factors Of Automobile Price Based On Random Forest And Neural Network

With the changes of the times and the rapid development of science and technology,people's material life has become more diverse than before.The increase in the number of private cars on the road is part of the diversification of this material life.In the past,cars were only owned by the rich.However,now every household is popularizing cars,and even some families have multiple cars.It can be seen that cars have gradually become the ideal transportation means in the minds of modern people.However,in the process of car selection and purchase,the price of the car will first affect people's final purchase situation.But it seems that many people don't know much about the relevant parameters in the car.Except for the brand and the space of the car(SUV,ordinary car,etc.),most of the index parameters may be outside the scope of people's involvement,and this will bring them some problems in the process of buying a car.For example,their psychological price does not match the model they like.Therefore,in order to more accurately understand the fluctuation factors of car prices,make it easier and more effective for consumers to buy the models they want,we need to analyze and predict the price of the car.Research on automobile prices,with a view to building a more accurate and more applicable automobile price model,to solve some problems that will occur when consumers buy cars as well as provide them with useful help or reference.The research in this paper firstly preprocesses the obtained car price data set appropriately,to make the data set reach the level that python can analyze and process.By deleting records and fields,encoding features,normalizing features,filling missing values and other processing methods,we achieved the purpose of cleaning and construction of the data set.In order to prevent overfitting,the principal component analysis algorithm is also used when processing multiple regression models.In terms of car price model selection,this paper uses three models including multiple linear regression,random forest regression,and fully connected neural network to process the data and compare them experimentally to obtain a better car price model.The innovation of this article is mainly reflected in:1.The random forest algorithm uses a combination of random search and grid search to adjust parameters,which is more robust than a single parameter adjustment method.2.Full connection neural network uses a completely new active function called"selu "function,which is more robust than other activation functions.