def _execute():
env = Environment(
train_dataset=get_breast_cancer_data(),
root_results_path='HyperparameterHunterAssets',
target_column='diagnosis',
metrics_map=['roc_auc_score'],
cross_validation_type=StratifiedKFold,
cross_validation_params=dict(n_splits=10,
shuffle=True,
random_state=32),
runs=1,
)
optimizer = RandomForestOptimization(
iterations=100,
read_experiments=True,
)
optimizer.set_experiment_guidelines(
model_initializer=LGBMClassifier,
model_init_params=dict(boosting_type=Categorical(['gbdt', 'dart']),
num_leaves=Integer(5, 20),
max_depth=-1,
min_child_samples=5,
subsample=0.5),
)
optimizer.go()
def opt_xgb_0():
optimizer = RandomForestOptimization(iterations=2, random_state=1337)
optimizer.set_experiment_guidelines(
model_initializer=XGBClassifier,
model_init_params=dict(
objective="reg:linear",
max_depth=Integer(2, 20),
learning_rate=Real(0.0001, 0.5),
subsample=0.5,
booster=Categorical(["gbtree", "dart"]),
),
)
optimizer.go()
yield optimizer
experiment_4 = CVExperiment(DecisionTreeClassifier, {})
experiment_5 = CVExperiment(RandomForestClassifier, {})
experiment_6 = CVExperiment(AdaBoostClassifier, {})
experiment_7 = CVExperiment(GradientBoostingClassifier, {})
experiment_8 = CVExperiment(GaussianNB, {})
experiment_9 = CVExperiment(LinearDiscriminantAnalysis, {})
experiment_10 = CVExperiment(QuadraticDiscriminantAnalysis, {})
experiment_11 = CVExperiment(MLPClassifier, {})
# Of course, SKLearn has many more algorithms than those shown here, but I think you get the idea
# Notice that in all the above experiments, we gave `CVExperiment` `model_init_params={}`.
# Passing an empty dict tells it to use the default hyperparameters for the `model_initializer`, which it'll figure out on its own.
#################### 2. Hyperparameter Optimization ####################
# We're just going to do optimization on one of the algorithms used above (`AdaBoostClassifier`);
# ... although, HyperparameterHunter can certainly do consecutive optimization rounds.
# Notice below that `optimizer` correctly identifies `experiment_6` as being the only saved
# ... experiment it can learn from because it's optimizing `AdaBoostClassifier`.
optimizer = RandomForestOptimization(iterations=12, random_state=42)
optimizer.set_experiment_guidelines(
model_initializer=AdaBoostClassifier,
model_init_params=dict(
n_estimators=Integer(25, 100),
learning_rate=Real(0.5, 1.0),
algorithm=Categorical(["SAMME", "SAMME.R"]),
),
)
optimizer.go()
cross_validation_params=dict(n_splits=5, random_state=32),
)
# Now that HyperparameterHunter has an active `Environment`, we can do two things:
#################### 1. Perform Experiments ####################
experiment = CVExperiment(
model_initializer=LGBMClassifier,
model_init_params=dict(boosting_type="gbdt",
num_leaves=31,
max_depth=-1,
subsample=0.5),
)
# And/or...
#################### 2. Hyperparameter Optimization ####################
optimizer = RandomForestOptimization(iterations=10, random_state=32)
optimizer.set_experiment_guidelines(
model_initializer=LGBMClassifier,
model_init_params=dict(
boosting_type=Categorical(["gbdt", "dart"]),
num_leaves=Integer(10, 40),
max_depth=-1,
subsample=Real(0.3, 0.7),
),
)
optimizer.go()
# Notice, `optimizer` recognizes our earlier `experiment`'s hyperparameters fit inside the search
# space/guidelines set for `optimizer`.
# Then, when optimization is started, it automatically learns from `experiment`'s results
# Now that HyperparameterHunter has an active `Environment`, we can do two things:
#################### 1. Perform Experiments ####################
experiment = CVExperiment(
model_initializer=LGBMClassifier,
model_init_params=dict(boosting_type="gbdt", num_leaves=31, max_depth=-1, subsample=0.5),
model_extra_params=dict(
fit=dict(
feature_name=train_df.columns.values[:-1].tolist(),
categorical_feature=train_df.columns.values[11:-1].tolist(),
)
),
)
# And/or...
#################### 2. Hyperparameter Optimization ####################
optimizer = RandomForestOptimization(iterations=10, random_state=32)
optimizer.set_experiment_guidelines(
model_initializer=LGBMClassifier,
model_init_params=dict(
boosting_type=Categorical(["gbdt", "dart"]),
num_leaves=Integer(10, 40),
max_depth=-1,
subsample=Real(0.3, 0.7),
),
model_extra_params=dict(
fit=dict(
feature_name=train_df.columns.values[:-1].tolist(),
categorical_feature=train_df.columns.values[11:-1].tolist(),
)
),
)