In data analysis and predictive modelling, in order to solve the problem that traditional point prediction methods cannot portray the uncertainty of predicted values, the article proposes an interval prediction model based on multiple regression analysis and Lasso-logistic. The model combines the numerical prediction capability of multiple regression with the feature screening and classification advantages of Lasso-Logistic to establish an interval prediction framework. The model is validated by examples, and it can effectively quantify the fluctuation range of the predicted values, improve the accuracy and robustness of the model, and is of great value in multi-disciplinary prediction.