Using sklearn to cross-validate bayesian network classifier

The purpose of this notebook is to show the possible integration of the pyAgrum’s classifier in the scikit-learn’s ecosystem. Thus, it is possible to use the tools provided by scikit-learn for crossfolding for pyAgrum’s Bayesian network.

from pyagrum.skbn import BNClassifier

from sklearn.model_selection import cross_validate
from sklearn import datasets

## get iris data
iris = datasets.load_iris()
X = iris.data
y = iris.target

model = BNClassifier(
  learningMethod="MIIC",
  prior="Smoothing",
  priorWeight=1,
  discretizationNbBins=3,
  discretizationStrategy="kmeans",
  discretizationThreshold=10,
)

cv = cross_validate(model, X, y, cv=30)
print(f"scores with cross-folding : {cv['test_score']}")
print()

print(f"mean score : {cv['test_score'].mean()}")

scores with cross-folding : [1.  1.  1.  1.  1.  1.  1.  1.  1.  1.  1.  1.  1.  1.  0.8 0.8 0.8 1.
 1.  1.  1.  0.8 1.  0.8 1.  1.  1.  1.  1.  1. ]

mean score : 0.9666666666666667

cv = cross_validate(model, X, y, cv=50)
print(f"scores with cross-folding : {cv['test_score']}")
print()

print(f"mean score : {cv['test_score'].mean()}")

scores with cross-folding : [1.         1.         1.         1.         1.         1.
 1.         1.         1.         1.         1.         1.
 1.         1.         1.         1.         1.         1.
 1.         1.         1.         1.         1.         0.66666667
 1.         0.66666667 1.         0.66666667 1.         1.
 1.         1.         1.         1.         1.         1.
 1.         1.         1.         0.66666667 1.         1.
 1.         1.         1.         1.         1.         1.
 1.         1.        ]

mean score : 0.9733333333333333