JOHNKYON / Sora

Visualization toolkit for ML and DL methods

Sora

Visualization toolkit for ML and DL methods

---

Version 1.0.0

New features:

• Support calculating SHAP values for scikit-learn decision tree regressor.
• Sample script to show how the result would look like.

How to calculate it?

1. Clone or download this repository and go into the root of the toolkit.
2. In your code, do as follows:

   from SHAP.tree import TreeExplainer
   
   # model is your own sklearn.tree.DecisionTreeRegressor()
   # Predictors: labels of x
   explainer = TreeExplainer(model).shap_values(x=data[0])
   print(explainer[0, :])
   
   plt.bar(range(len(predictors)), explainer[0, :-1], tick_label=predictors)

How to run the sample script:

1. Clone or download this repository and go into the root of the toolkit.
2. Run $ python sample_script.tree.decision_tree_sample.py

Then you are likely to see a shap.png in the pwd. The picture is a bar figure and should look like