Analysis And Research On Price Forecasting Model Of Electronic Products

With the globalization of the economy and the continuous development of science and technology,product price prediction plays an important role in the field of economic finance and product production.Because of the many attributes of the product,predicting the price of the product requires selecting the appropriate product characteristics and the method of prediction.This thesis mainly studies the price forecasting model from the aspects of random forest model,ridge regression model,XGBoost model and the combination models.The main contents are as follows:Preprocess the original data,due to some incomplete or inconsistent data of the original data,and the existence of discrete dataset features with non-real number value description,the data preprocessing technique is used to process the original data and prepare data for subsequent model training.According to the importance of the random forest to measure the characteristics of the data,the feature items with higher importance are selected as the data features needed for the model training to improve the accuracy of the model prediction.The linear regression algorithm,the ridge regression algorithm,the decision tree algorithm,the random forest algorithm and the XGBoost algorithm are studied.The electronic product price data set is used to analyze the above algorithms,and the prediction algorithm with better prediction effect is selected.Because there is a price difference in the data with similar characteristics in the dataset,the clustering analysis method is first used to obtain a new training dataset,and the price forecasting model is designed based on the new training dataset,and the single model is adjusted and the single model under different parameter values is compared with the root mean square error to determine the single model with the ideal prediction effect.The weight of each single model is obtained by the Boosting algorithm,and the single model is fused by the weighted average method to obtain the final combined model.Using the combined model to predict the price of electronic products,and comparing the root mean squared error,the mean absolute error and R-squared value of the single model and the combined model.The experimental results show that the combined model has better prediction effect,and the combined model's prediction results and actual price fit is higher.