The model of price of sailing ships based on Lasso regression

For the sample data of sailing ships and the listed price prediction of sailing ships based on the characteristics of sailing ships found on the website, we first conducted data cleaning on the original data obtained. In this stage, there were many missing values and outliers in the original data. After filling the missing values with mode, We transform the classified variables into dummy variables, and finally normalize them to convert the original data into the training data of the model. Then, we obtained the predicted value of Listing Price (USD) through multiple regression fitting. By calculating R2 as 0.929, it was found that the model fitting effect was perfect, but there were too many variables due to the conversion of attribute variables to dummy variables, so it was necessary to compress model variables to select key variables. Since this topic is the explanation of Listing Price (USD), the coefficient of each variable needs to be known, so tree model is not adopted. In linear model, Lasso regression mainly screens model variables. In this case, Lasso is the main screening method. The mean square error of the listing price predicted by the multiple regression model based on Lasso regression adjustment parameters is 0.125, indicating that the model has high accuracy and the simulated listing price predicted is relatively high.