def evaluate():
input_file = os.path.join("outputs", "features",
"function_and_human_readable_features.json")
X, y = preprocess.get_features_and_labels(input_file)
skf = StratifiedKFold(n_splits=5, random_state=42)
splits = list(skf.split(X, y))
svm_run(X, y, splits)
decision_tree_run(X, y, splits)
for loss in [
'hinge', 'log', 'modified_huber', 'squared_hinge', 'perceptron'
]:
sgd_run(X, y, splits, loss)
-1: 20
}, 'balanced']
}
n_iter_search = 100
random_search = RandomizedSearchCV(DecisionTreeClassifier(),
param_distributions=param_dist,
n_iter=n_iter_search,
scoring={
'precision': precision_scorer,
'recall': recall_scorer,
'f1': f1_scorer,
'geometric_mean': gmean_scorer
},
cv=cv,
refit='f1',
n_jobs=-1,
random_state=42)
random_search.fit(X, y)
return report(random_search.cv_results_)
if __name__ == "__main__":
print("-- Random Parameter Search via 4-fold CV")
input_file = os.path.join("outputs", "features",
"function_and_human_readable_features.json")
X, y = preprocess.get_features_and_labels(input_file)
_ = decision_tree_randomized_search(X, y)
