eval_Coxmos_model_per_variable
The eval_Coxmos_model_per_variable function offers a granular evaluation of a specific Coxmos model, focusing on the influence of individual variables or components on the model's predictive performance. It computes the Area Under the Curve (AUC) for each variable at designated time points, providing insights into the relative importance of each variable in the model's predictions. For a visual representation of the results, it is advisable to utilize the plot_evaluation()
function post-evaluation.
eval_Coxmos_model_per_variable( model, X_test, Y_test, pred.method = "cenROC", pred.attr = "mean", times = NULL, max_time_points = 15, PARALLEL = FALSE, verbose = FALSE )
model: Coxmos model.X_test: Numeric matrix or data.frame. Explanatory variables for test data (raw format). Qualitative variables must be transform into binary variables.Y_test: Numeric matrix or data.frame. Response variables for test data. Object must have two columns named as "time" and "event". For event column, accepted values are: 0/1 or FALSE/TRUE for censored and event observations.pred.method: Character. AUC evaluation algorithm method for evaluate the model performance. Must be one of the following: "risksetROC", "survivalROC", "cenROC", "nsROC", "smoothROCtime_C", "smoothROCtime_I" (default: "cenROC").pred.attr: Character. Way to evaluate the metric selected. Must be one of the following: "mean" or "median" (default: "mean").times: Numeric vector. Time points where the AUC will be evaluated. If NULL, a maximum of 'max_time_points' points will be selected equally distributed (default: NULL).max_time_points: Numeric. Maximum number of time points to use for evaluating the model (default: 15).PARALLEL: Logical. Run the cross validation with multicore option. As many cores as your total cores - 1 will be used. It could lead to higher RAM consumption (default: FALSE).verbose: Logical. If verbose = TRUE, extra messages could be displayed (default: FALSE).A list of two objects: df: A data.frame which the predictions for the specific model split into the full model (LP) and each component individually. This data.frame is used to plot the information by the function plot_evaluation(). lst_AUC: A list of the full model prediction and its components where the user can check the linear predictors used, the global AUC, the AUC per time point and the predicted time points selected.
Upon invocation, the function initiates by verifying the consistency between test times and the training times of the provided model. Subsequently, linear predictors for each variable are derived using the predict.Coxmos function. These linear predictors serve as the foundation for the AUC computation, which is executed for each variable across the specified time points.
The function employs various evaluation methods, as determined by the pred.method parameter, to calculate the AUC values. These methods encompass options such as "risksetROC", "survivalROC", and "cenROC", among others. The results are systematically organized into a structured data frame, segregating AUC values for each variable at different time points. This structured output not only facilitates easy interpretation but also sets the stage for subsequent visualization or further analysis.
It's noteworthy that the function is equipped to handle parallel processing, contingent on the user's preference, which can expedite the evaluation process, especially when dealing with extensive datasets or multiple time points.
data("X_proteomic") data("Y_proteomic") set.seed(123) index_train <- caret::createDataPartition(Y_proteomic$event, p = .5, list = FALSE, times = 1) X_train <- X_proteomic[index_train,1:50] Y_train <- Y_proteomic[index_train,] X_test <- X_proteomic[-index_train,1:50] Y_test <- Y_proteomic[-index_train,] model_icox <- splsicox(X_train, Y_train, n.comp = 2) eval_Coxmos_model_per_variable(model_icox, X_test, Y_test, pred.method = "cenROC")
Pedro Salguero Garcia. Maintainer: pedsalga@upv.edu.es