predictionStats: Prediction Evaluation

In FRESA.CAD: Feature Selection Algorithms for Computer Aided Diagnosis

Prediction Evaluation

Description

This function returns the statistical metrics describing the association between model predictions and the ground truth outcome

Usage

      predictionStats_binary(predictions, plotname="", center=FALSE,...)
      predictionStats_regression(predictions, plotname="",...)
      predictionStats_ordinal(predictions,plotname="",...)
      predictionStats_survival(predictions,plotname="",atriskthr=1.0,...)

Arguments

predictions	A matrix whose first column is the ground truth, and the second is the model prediction
plotname	The main title to be used by the plot function. If empty, no plot will be provided
center	For binary predictions indicates if the prediction is around zero
atriskthr	For survival predictions indicates the threshoold for at risk subjects.
...	Extra parameters to be passed to the plot function.

Details

These functions will analyze the prediction outputs and will compare to the ground truth. The output will depend on the prediction task: Binary classification, Linear Regression, Ordinal regression or Cox regression.

Value

accc	The classification accuracy with its95% confidence intervals (95/
berror	The balanced error rate with its 95%CI
aucs	The ROC area under the curve (ROC AUC) of the binary classifier with its 95%CI
specificity	The specificity with its 95%CI
sensitivity	The sensitivity with its 95%CI
ROC.analysis	The output of the ROC function
CM.analysis	The output of the epiR::epi.tests function
corci	the Pearson correlation with its 95%CI
biasci	the regression bias and its 95%CI
RMSEci	the root mean square error (RMSE) and its 95%CI
spearmanci	the Spearman correlation and its 95%CI
MAEci	the mean absolute difference(MAE) and its 95%CI
pearson	the output of the cor.test function
Kendall	the Kendall correlation and its 95%CI
Bias	the ordinal regression bias and its 95%CI
BMAE	the balanced mean absolute difference for ordinal regression
class95ci	the output of the bootstrapped estimation of accuracy, sensitivity, and ROC AUC
KendallTauB	the output of the DescTools::KendallTauB function
Kappa.analysis	the output of the irr::kappa2 function
CIFollowUp	The follow-up concordance index with its95% confidence intervals (95/
CIRisk	The risks concordance index with its95% confidence intervals (95/
LogRank	The LogRank test with its95% confidence intervals (95/

Author(s)

Jose G. Tamez-Pena

See Also

randomCV

FRESA.CAD documentation built on Nov. 25, 2023, 1:07 a.m.