Evaluate classification prediction results.
Usage
X.evaluate(
data = NULL,
scores.col = NA,
labels.col = NA,
eval.metric = "prc",
plot = FALSE,
black = FALSE,
kappa.cutoff = 0.5,
kappa.weight = "unweighted",
F1.cutoff = 0.5
)Arguments
- data
Data frame with columns for model scores and labels.
- scores.col
Character string: 'data' column name with probs that is to be evaluated.
- labels.col
Character string: 'data' column name with labels (1 for complex-forming, 0 for others.)
- eval.metric
Character string: Method for predictor evaluation. 'roc' for area under the receiver-operator curve, 'prc' for area under the precision-recall curve, 'kappa' for Cohen's kappa and 'F1' for F1 score.
- plot
Logical: Should the plots be plotted and outputted?
- black
Logical: Should the plotted curve be black or color gradient? Default is FALSE.
- kappa.cutoff
Numeric: What is the score cutoff for evaluating based on Cohen's kappa? Default is 0.5.
- kappa.weight
Character string: Same as argument weight in irr:kappa2
- F1.cutoff
Numeric: What is the score cutoff for evaluating based on F1? Default is 0.5.