threshold_misclass: Threshold misclassification
In messamat/globalIRmap: What the Package Does (One Line, Title Case)

Description Usage Arguments Details Value Warning

Compute the misclassification rate, sensitivity, and specificity of probability predictions to a binary classification problem from a random forest learner based on a given threshold.

1	threshold_misclass(i = 0.5, in_preds)

i

(numeric) Threshold to be used for converting probabilities predictions to binary predictions, limits=[0,1].

in_preds

Either:

a PredictionClassif or
a data.table of predictions for a set of CV repetitions as formatted by analyze_benchmark. The column of the predicted probability of getting a 'positive' must be called prob.1.

Misclassification rate is the proportion of misclassified records
Sensitivity is the proportion of actual positives (1) that are correctly identified as such (the proportion of intermittent rivers that are identified as intermittent)
Specificity is the proportion of actual negatives that are correctly identified as such (the proportion of perennial rivers that are identified as perennial)

See https://en.wikipedia.org/wiki/Sensitivity_and_specificity

A single-row data.table with four columns:

i - the threshold used to compute the classification statistics
misclass - misclassification rate [0-1]
sens - sensitivity [0-1]
spec - specificity [0-1]

This function was only tested for the outputs of a probability random forest (using a classification framework) and a regression forest (using 0 or 1 as dummy continuous variables)

messamat/globalIRmap documentation built on July 4, 2021, 10:48 a.m.