2.2.3 Construction and evaluation of predictive models
In this study, Stacking Ensemble Learning was used to predict the level of TNF-α. The 4 basic models included in the stacking ensemble learning are as follows: Random Forest (RF), Extreme Gradient Boosting (XGBoost) model, Least Absolute Shrinkage and Selection Operator (Lasso), Single Hidden Layer Neural Network (SHLNN). We use data standardization to optimize the data, and use v-fold for 5-fold cross-validation (v =5), and each fold performs 10 random divisions (break =10) as a resampling method. Finally, set up the model and use the grid search method to optimize the hyperparameters of the corresponding model, and use cross-validation to evaluate the model performance of the corresponding parameters. Afterwards, the constructed model is added to the Stacking Ensemble Model as a meta-feature, and lasso regression is used to fit the meta-model features to obtain a better model, and then the model is used to predict and evaluate the training set and the test set respectively. At the same time, use Root Mean Squared Error (RMSE), R-squared (R²) and Mean Absolute Error (MAE) as evaluation indicators to evaluate the Stacking Ensemble Model.