1. Black-box interpretation of models: LIME

See also skater: a more modern variant relying on heavier dependencies https://github.com/datascienceinc/Skater

First we need to install lime with the following shell line:

$ pip install --user lime

Lime uses the notion of “explainers”, for different types of data, tabular, image, text.

1.1. Regression on tabular data: factors of prices of houses

Load the data, create and fit a regressor

from sklearn import datasets, ensemble, model_selection

boston = datasets.load_boston()
X_train, X_test, y_train, y_test = model_selection.train_test_split(
    boston.data, boston.target)

regressor = ensemble.RandomForestRegressor()
regressor.fit(X_train, y_train)

Inspect predictions for a few houses

For this, separate out the categorical features:

import numpy as np
categorical_features = [i for i, col in enumerate(boston.data.T)
                        if np.unique(col).size < 10]

Now use a lime explainer for tabular data

from lime.lime_tabular import LimeTabularExplainer
explainer = LimeTabularExplainer(X_train,
    feature_names=boston.feature_names,
    class_names=['price'],
    categorical_features=categorical_features,
    mode='regression')

# Now explain a prediction
exp = explainer.explain_instance(X_test[25], regressor.predict,
        num_features=10)

exp.as_pyplot_figure()
from matplotlib import pyplot as plt
plt.tight_layout()

print(exp.as_list())

Out:

[('LSTAT <= 7.44', 7.7596752211035458), ('6.18 < RM <= 6.58', -0.9093968629223903), ('DIS > 4.72', -0.70512543216415358), ('NOX <= 0.46', 0.51560566507743), ('CRIM <= 0.09', -0.45722319351284785), ('ZN > 0.00', 0.39788175678697063), ('B > 396.00', 0.3814292667526803), ('19.10 < PTRATIO <= 20.20', -0.38084477008612727), ('RAD=1', -0.30284784526829844), ('281.00 < TAX <= 345.00', -0.30202335699846755)]

Explain a few more predictions

for i in [7, 50, 66]:
    exp = explainer.explain_instance(X_test[i], regressor.predict,
            num_features=10)
    exp.as_pyplot_figure()
    plt.tight_layout()

• 
• 
• 

Total running time of the script: ( 0 minutes 1.539 seconds)