d3:evalpred {BCBCSF}R Documentation

A function for evaluating arrays of predictive probabilities with the true class labels of test cases

Description

This function is used to find error rate, amlp, loss and predictive probabilities at true labels.

Usage

eval_pred (out_pred, y_ts, Mloss = NULL)

Arguments

out_pred

a list returned by function bcbcsf_fitpred with X_ts given, or bcbcsf_pred, or by cross_vld.

y_ts

a vector of true class labels.

Mloss

a matrix indicting loss function, with element m_{ij} saving the losss from predicting class i with class label j

; by default, it is NULL.

Value

probs_at_truelabels

a matrix of predictive probabilities at true labels, with rows for cases, and columns for different numbers of retained features

summary

a data frame, with rows for different numbers of retained features, and columns: Error.Rate: fraction of cases misclassified with fair threshold, and AMLP: minus average log probabilities at true labels, often called "deviation", and Loss (if Mloss is given): average loss.


[Package BCBCSF version 1.0-1 Index]