from skopt.space import Integer, Real
from sklearn.pipeline import Pipeline
from utility import HyperParameters, Runner
from model import load_sample_data_frame, ordinal_data_mapper
sample = None
iterations = 24
hyper_parameters = HyperParameters(
search_space={
'lgb__n_estimators': Integer(50, 500),
'lgb__learning_rate': Real(0.001, 0.05),
'lgb__num_leaves': Integer(7, 63),
'lgb__max_bin': Integer(32, 256),
'lgb__min_child_samples': Integer(10, 50),
'lgb__subsample': Real(0.5, 0.9),
'lgb__colsample_bytree': Real(0.5, 0.9)
})
lightgbm_basic = Pipeline([('mapper', ordinal_data_mapper),
('lgb', lgb.LGBMClassifier())])
def test_lightgbm_basic():
runner = Runner('model/experiment/output/lightgbm_basic',
load_sample_data_frame(), 'violation', lightgbm_basic,
hyper_parameters)
runner.run_classification_search_experiment('neg_log_loss',
sample=sample,
from skopt.space import Categorical, Integer, Real
from sklearn.pipeline import Pipeline
from sklearn.tree import DecisionTreeClassifier
from utility import HyperParameters, Runner
from model import load_clean_sample_data_frame, ordinal_data_mapper
sample = None
iterations = 24
hyper_parameters = HyperParameters({
'dt__criterion':
Categorical(['gini', 'entropy']),
'dt__max_depth':
Integer(4, 24),
'dt__min_samples_leaf':
Real(0.000001, 0.001),
'dt__min_samples_split':
Real(0.000002, 0.002)
})
decision_tree_basic = Pipeline([('mapper', ordinal_data_mapper),
('dt', DecisionTreeClassifier())])
def test_decision_tree():
runner = Runner('model/experiment/output/decision_tree_basic',
load_clean_sample_data_frame(),
'arrest',
decision_tree_basic,
hyper_parameters=hyper_parameters)
from imblearn.under_sampling import RandomUnderSampler
from skopt.space import Categorical, Integer, Real
from sklearn.ensemble import GradientBoostingClassifier
from sklearn.pipeline import Pipeline
from utility import HyperParameters, Runner
from model import load_clean_sample_data_frame, ordinal_data_mapper
sample = None
iterations = 12
hyper_parameters = HyperParameters({
'gb__learning_rate': Real(0.01, 0.1),
'gb__subsample': Real(0.5, 1),
'gb__max_depth': Integer(3, 7),
'gb__max_features': Categorical(['sqrt', 'log2'])
})
gradient_boosting_pipeline = Pipeline([
('mapper', ordinal_data_mapper),
('gb', GradientBoostingClassifier(n_estimators=200))
])
def test_gradient_boosting():
runner = Runner(
'model/experiment/output/gradient_boosting_basic',
load_clean_sample_data_frame(),
'arrest',
gradient_boosting_pipeline,
from skopt.space import Integer, Real
from sklearn.pipeline import Pipeline
from utility import HyperParameters, Runner
from model import load_sample_data_frame, ordinal_data_mapper
sample = None
iterations = 24
hyper_parameters = HyperParameters(
search_space={
'xgb__n_estimators': Integer(100, 500),
'xgb__learning_rate': Real(0.1, 0.3),
'xgb__gamma': Real(0.0001, 100.0, prior='log-uniform'),
'xgb__max_depth': Integer(3, 7),
'xgb__colsample_bytree': Real(0.4, 0.8),
'xgb__colsample_bylevel': Real(0.4, 0.8),
'xgb__colsample_bynode': Real(0.4, 0.8)
})
xgboost_basic = Pipeline([('mapper', ordinal_data_mapper),
('xgb', xgb.XGBClassifier(tree_method='hist'))])
def test_xgboost():
runner = Runner('model/experiment/output/xgboost_basic_full',
load_sample_data_frame(),
'arrest',
xgboost_basic,
hyper_parameters=hyper_parameters)
from skopt.space import Categorical, Integer
from sklearn.ensemble import RandomForestClassifier
from sklearn.pipeline import Pipeline
from utility import HyperParameters, Runner
from model import load_sample_data_frame, ordinal_data_mapper
sample = None
iterations = 24
hyper_parameters = HyperParameters(search_space={
'rf__n_estimators': Integer(50, 100),
'rf__criterion': Categorical(['gini', 'entropy']),
'rf__max_depth': Integer(4, 18),
'rf__max_features': Categorical(['sqrt', 'log2', None]),
'rf__bootstrap': Categorical([True, False])
})
random_forest_basic = Pipeline([
('mapper', ordinal_data_mapper),
('rf', RandomForestClassifier())
])
def test_random_forest_basic():
runner = Runner(
'model/experiment/output/random_forest_basic',
load_sample_data_frame(),
'violation',
from skopt.space import Categorical, Integer, Real
from sklearn.ensemble import AdaBoostClassifier
from sklearn.pipeline import Pipeline
from utility import HyperParameters, Runner
from model import load_clean_sample_data_frame, ordinal_data_mapper
sample = None
iterations = 12
hyper_parameters = HyperParameters({
'ada__n_estimators':
Integer(50, 200),
'ada__learning_rate':
Real(0.5, 1.5),
'ada__algorithm':
Categorical(['SAMME', 'SAMME.R'])
})
ada_boost_pipeline = Pipeline([('mapper', ordinal_data_mapper),
('ada', AdaBoostClassifier())])
def test_ada_boost():
runner = Runner('model/experiment/output/ada_boost_basic',
load_clean_sample_data_frame(),
'arrest',
ada_boost_pipeline,
hyper_parameters=hyper_parameters)
runner.run_classification_search_experiment('roc_auc',
from skopt.space import Categorical, Integer, Real
from sklearn.ensemble import RandomForestClassifier
from sklearn.pipeline import Pipeline
from utility import HyperParameters, Runner
from model import load_clean_sample_data_frame, ordinal_data_mapper
sample = None
iterations = 12
hyper_parameters = HyperParameters({
'rf__n_estimators':
Integer(50, 150),
'rf__criterion':
Categorical(['gini', 'entropy']),
'rf__max_depth':
Integer(4, 18),
'rf__max_features':
Categorical(['sqrt', 'log2'])
})
random_forest_pipeline = Pipeline([('mapper', ordinal_data_mapper),
('rf', RandomForestClassifier())])
def test_random_forest():
runner = Runner('model/experiment/output/random_forest_basic',
load_clean_sample_data_frame(),
'arrest',
random_forest_pipeline,
hyper_parameters=hyper_parameters)