InterpretMe.jl

A Machine Learning Interpretability Library for Julia
Author mmahrouss
Popularity
1 Star
Updated Last
4 Years Ago
Started In
October 2019

InterpretMe

Build Status Build Status Codecov Coveralls

InterpretMe is Julia Package aimed at providing interpret-ability tools for Machine Learning models in Julia.

Installation

julia>]
pkg> add InterpretMe

Example

First import the package:

Using InterpretMe

Then train a model from ScikitLearn.jl's examples. We are going to use the Iris Example

Load the packages and train the model:

import ScikitLearn
import RDatasets

ScikitLearn.@sk_import linear_model: LogisticRegression

iris = RDatasets.dataset("datasets", "iris")

X = convert(Array, iris[[:SepalLength, :SepalWidth, :PetalLength, :PetalWidth]])
y = convert(Array, iris[:Species])

model = ScikitLearn.fit!(LogisticRegression(), X, y)

Now we can calculate the partial_dependance function from InterpretMe.

pd, vals = partial_dependence(x, [1], x -> ScikitLearn.predict_proba(model,x),
                              100, false)