Data-Driven Prediction Based on Gradient Boosted Regression Trees in a Machine Learning Framework

This paper focuses on the application of gradient boosted regression trees in data processing and predictive analytics in the computer field. Firstly, for large dataset processing, the XGBoost algorithm is applied, and hyperparameter optimization is performed with the help of Optuna, taking the mean square error as the optimization objective function, and at the same time, error indicators such as MSE, RMSE and MAE are used to evaluate the model effect. Secondly, the Lazypredict automated model selection framework was utilized to compare the performance of multiple supervised learning models, and the gradient boosted regression tree was identified as the better model after comprehensively considering the indicators of R-Squared, adjusted R-Squared and RMSE. Finally, the gradient boosting regression tree algorithm is used to construct the model, select specific information as features, set relevant data as target variables, and evaluate the model again by error indicators such as MSE, RMSE, and MAE, so as to effectively explore the potential relationship between the data, and to provide powerful support for decision-making in related fields.