def test_dtc(breast_cancer_data, caplog):
import logging
from generalizedtrees.recipes import binary_decision_tree_classifier
from generalizedtrees.features import FeatureSpec
logger = logging.getLogger()
caplog.set_level(logging.DEBUG)
logger.info("Creating class instance")
dtc = binary_decision_tree_classifier(max_depth=5)
logger.info("Fitting tree")
d = breast_cancer_data.x_train.shape[1]
dtc.fit(breast_cancer_data.x_train,
breast_cancer_data.y_train,
feature_spec=(FeatureSpec.CONTINUOUS, ) * d)
logger.info(f'Learned tree:\n{dtc.show_tree()}')
logger.info("Running prediction")
dtc.predict(breast_cancer_data.x_test)
logger.info("Done")
def test_dtc_json(breast_cancer_data_pandas, caplog):
import logging
import pandas as pd
from generalizedtrees.recipes import binary_decision_tree_classifier
from generalizedtrees.vis import explanation_to_JSON
from generalizedtrees.vis.vis import explanation_to_simplified
logger = logging.getLogger()
caplog.set_level(logging.DEBUG)
logger.info("Creating class instance")
dtc = binary_decision_tree_classifier(max_depth=5)
logger.info("Fitting tree")
dtc.fit(breast_cancer_data_pandas.x_train,
breast_cancer_data_pandas.y_train)
logger.info(f'Learned tree:\n{dtc.show_tree()}')
annotation = pd.DataFrame(
{'names': breast_cancer_data_pandas.x_train.columns})
logger.info(f'Simplified: {explanation_to_simplified(dtc, annotation)}')
logger.info(f'JSON: {explanation_to_JSON(dtc, annotation)}')
def test_dtc_bin_vs_multi(breast_cancer_data_pandas, caplog):
import logging
from generalizedtrees.recipes import binary_decision_tree_classifier, decision_tree_classifier
logger = logging.getLogger()
caplog.set_level(logging.DEBUG)
logger.info("Creating binary version")
dtc_bin = binary_decision_tree_classifier(max_depth=5)
dtc_bin.fit(breast_cancer_data_pandas.x_train,
breast_cancer_data_pandas.y_train)
logger.info(f'Learned tree:\n{dtc_bin.show_tree()}')
logger.info("Creating general version")
dtc_gen = decision_tree_classifier(max_depth=5)
dtc_gen.fit(breast_cancer_data_pandas.x_train,
breast_cancer_data_pandas.y_train)
logger.info(f'Learned tree:\n{dtc_gen.show_tree()}')
logger.info('Running predictions')
bin_prob = dtc_bin.predict_proba(breast_cancer_data_pandas.x_test)
bin_lab = dtc_bin.predict(breast_cancer_data_pandas.x_test)
gen_prob = dtc_gen.predict_proba(breast_cancer_data_pandas.x_test)
gen_lab = dtc_gen.predict(breast_cancer_data_pandas.x_test)
logger.info("Comparing")
assert_allclose(bin_prob, gen_prob)
assert_allclose(bin_lab, gen_lab)
logger.info("Done")
def test_dtc_pandas(breast_cancer_data_pandas, caplog):
import logging
from generalizedtrees.recipes import binary_decision_tree_classifier
logger = logging.getLogger()
caplog.set_level(logging.DEBUG)
logger.info("Creating class instance")
dtc = binary_decision_tree_classifier(max_depth=5)
logger.info("Fitting tree")
dtc.fit(breast_cancer_data_pandas.x_train,
breast_cancer_data_pandas.y_train)
logger.info(f'Learned tree:\n{dtc.show_tree()}')
logger.info("Running prediction")
dtc.predict(breast_cancer_data_pandas.x_test)
logger.info("Done")
def test_dtc_serialization(breast_cancer_data, caplog):
import logging
from generalizedtrees.recipes import binary_decision_tree_classifier
from generalizedtrees.features import FeatureSpec
from pickle import dumps, loads
logger = logging.getLogger()
caplog.set_level(logging.DEBUG)
logger.info("Creating class instance")
dtc = binary_decision_tree_classifier(max_depth=5)
logger.info("Fitting tree")
d = breast_cancer_data.x_train.shape[1]
dtc.fit(breast_cancer_data.x_train,
breast_cancer_data.y_train,
feature_spec=(FeatureSpec.CONTINUOUS, ) * d)
tree_str = dtc.show_tree()
logger.info(f'Learned tree:\n{tree_str}')
logger.info('Pickling tree')
bytes_obj = dumps(dtc)
logger.info('Unpickling tree')
returned_dtc = loads(bytes_obj)
returned_tree_str = returned_dtc.show_tree()
logger.info(f'Unpickled tree:\n{returned_tree_str}')
assert returned_tree_str == tree_str
logger.info("Running unpicled tree prediction")
returned_dtc.predict(breast_cancer_data.x_test)
logger.info("Done")