davek44/metrics

Torch metrics package

Torch metrics package

This package provides utility functions to evaluate your machine learning models.

Receiver Operator Curves (ROC)

Used to evalute performance of binary classifiers, and their trade-offs in terms of false-positive and false-negative rates.

require 'torch'
metrics = require 'metrics'
gfx = require 'gfx.js'

resp = torch.DoubleTensor { -0.9, -0.8, -0.8, -0.5, -0.1, 0.0, 0.2, 0.2, 0.51, 0.74, 0.89}
labels = torch.IntTensor  { -1, -1 , 1, -1, -1, 1, 1, -1, -1, 1, 1 }

roc_points = metrics.ROC.points(resp, labels)
area = metrics.ROC.area(roc_points)

print(roc_points)
print(area)

gfx.chart(roc_points)

Confusion matrix (TODO)

Used to evaluate performance of multi-class classifiers.