MLTools.jl

Utilities for Machine Learning in Julia

To use, run the command:

Pkg.clone("https://github.com/Wedg/MLTools.jl.git")

ROC Curve

The ROC plot is built with the function:

• plot_ROC_curve(y_cond, y_prob)

  y_cond is the "truth" vector with each element either 0 or 1.
  y_prob is the hypothesis vector with each element a probability in the range [0, 1].

As well as the ROC curve of the predictor, the plot shows the model's accuracy, true positive rate, and false positive rate as well as the summary statistic AUC (Area Under Curve).

The plot will look something like this.

Confusion Matrix

The confusion matrix plot is built with the function:

• plot_confusion_matrix(y_cond, y_pred, "y_label", "x_label", classes)

  y_cond is the "truth" vector with each element one of the labels 1 through k.
  y_pred is the model prediction vector with each element one of the labels 1 through k.
  "y_label" is the string that will be the label of the y axis.
  "x_label" is the string that will be the label of the x axis.
  classes is a vector of strings representing the classes.

An example will hopefully make this clearer. This is taken from the Alice package demo of a reduced STL10 dataset where the classes are one of airplane, car, cat or dog.

See the demo for more context. The plot is called with the following:

classes = ["airplane", "car", "cat", "dog"]
plot_confusion_matrix(y_cond, y_pred, "Truth", "Prediction", classes)

And produces this plot