Global Explainability

Installation

# To install the required package, use the following command:
# !pip install modeva

Authentication

# To get authentication, use the following command: (To get full access please replace the token to your own token)
# from modeva.utils.authenticate import authenticate
# authenticate(auth_code='eaaa4301-b140-484c-8e93-f9f633c8bacb')

Import modeva modules

from modeva import DataSet
from modeva import TestSuite
from modeva.models import MoLGBMRegressor

Load Dataset

ds = DataSet()
ds.load(name="BikeSharing")
ds.set_random_split()

ds.scale_numerical(features=("cnt",), method="log1p")
ds.preprocess()

Train a LGBM model

model = MoLGBMRegressor(verbose=-1)
model.fit(ds.train_x, ds.train_y)

MoLGBMRegressor(boosting_type='gbdt', class_weight=None, colsample_bytree=1.0,
                importance_type='split', learning_rate=0.1, max_depth=-1,
                min_child_samples=20, min_child_weight=0.001,
                min_split_gain=0.0, n_estimators=100, n_jobs=None,
                num_leaves=31, objective=None, random_state=None, reg_alpha=0.0,
                reg_lambda=0.0, subsample=1.0, subsample_for_bin=200000,
                subsample_freq=0, verbose=-1)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

Permutation feature importance

ts = TestSuite(ds, model)
results = ts.explain_pfi(dataset='test', sample_size=2000, n_repeats=5, random_state=0)
results.plot(n_bars=10)

H-statistic

results = ts.explain_hstatistic(features=('hr',
                                          'atemp',
                                          'season',
                                          'holiday',
                                          'hum'),
                                dataset='train', sample_size=2000, percentiles=(0, 1),
                                grid_resolution=10, response_method='auto', random_state=0)
results.table

	hr	atemp	season	holiday	hum
hr	NaN	0.0003	0.0005	0.0010	0.0031
atemp	0.0003	NaN	0.0052	0.0448	0.0083
season	0.0005	0.0052	NaN	0.0029	0.0116
holiday	0.0010	0.0448	0.0029	NaN	0.0036
hum	0.0031	0.0083	0.0116	0.0036	NaN

1D Partial dependency plots

results = ts.explain_pdp(features="hr", dataset='train', sample_size=2000, percentiles=(0, 1),
                         grid_resolution=10, response_method='auto', random_state=0)
results.plot()

2D Partial dependency plots

results = ts.explain_pdp(features=("hum", "hr"), dataset="train")
results.plot()

1D ALE

results = ts.explain_ale(features="hr", dataset='train', sample_size=2000,
                         grid_resolution=10, response_method='auto', random_state=0)
results.plot()

2D ALE

results = ts.explain_ale(features=("hum", "hr"), dataset="train")
results.plot()