Models#

Different base obj choices for the Model are shown below The exact str indicator, as passed to the obj param is represented by the sub-heading (within “”) The available models are further broken down by which can work with different problem_types. Additionally, a link to the original models documentation as well as the implemented parameter distributions are shown.

binary#

“dt classifier”#

Base Class Documentation: sklearn.tree.DecisionTreeClassifier

Param Distributions

  1. “default”

    defaults only

  2. “dt classifier dist”

    max_depth: Scalar(lower=1, upper=30).set_mutation(sigma=4.833333333333333).set_bounds(full_range_sampling=True, lower=1, upper=30).set_integer_casting()
min_samples_split: Scalar(lower=2, upper=50).set_mutation(sigma=8.0).set_bounds(full_range_sampling=True, lower=2, upper=50).set_integer_casting()
class_weight: TransitionChoice([None, 'balanced'])

“elastic net logistic”#

Base Class Documentation: sklearn.linear_model.LogisticRegression

Param Distributions

  1. “base elastic”

    max_iter: 100
multi_class: 'auto'
penalty: 'elasticnet'
class_weight: None
solver: 'saga'
l1_ratio: 0.5

  2. “elastic classifier”

    max_iter: 100
multi_class: 'auto'
penalty: 'elasticnet'
class_weight: TransitionChoice([None, 'balanced'])
solver: 'saga'
l1_ratio: Scalar(lower=0.01, upper=1).set_mutation(sigma=0.165).set_bounds(full_range_sampling=True, lower=0.01, upper=1)
C: Log(lower=1e-05, upper=100000.0)

  3. “elastic clf v2”

    max_iter: 100
multi_class: 'auto'
penalty: 'elasticnet'
class_weight: TransitionChoice([None, 'balanced'])
solver: 'saga'
l1_ratio: Scalar(lower=0.01, upper=1).set_mutation(sigma=0.165).set_bounds(full_range_sampling=True, lower=0.01, upper=1)
C: Log(lower=0.01, upper=100000.0)

  4. “elastic classifier extra”

    max_iter: Scalar(lower=100, upper=1000).set_mutation(sigma=150.0).set_bounds(full_range_sampling=True, lower=100, upper=1000).set_integer_casting()
multi_class: 'auto'
penalty: 'elasticnet'
class_weight: TransitionChoice([None, 'balanced'])
solver: 'saga'
l1_ratio: Scalar(lower=0.01, upper=1).set_mutation(sigma=0.165).set_bounds(full_range_sampling=True, lower=0.01, upper=1)
C: Log(lower=1e-05, upper=100000.0)
tol: Log(lower=1e-06, upper=0.01)

“et classifier”#

Base Class Documentation: sklearn.ensemble.ExtraTreesClassifier

Param Distributions

  1. “default”

    defaults only

“gaussian nb”#

Base Class Documentation: sklearn.naive_bayes.GaussianNB

Param Distributions

  1. “base gnb”

    var_smoothing: 1e-09

“gb classifier”#

Base Class Documentation: sklearn.ensemble.GradientBoostingClassifier

Param Distributions

  1. “default”

    defaults only

“gp classifier”#

Base Class Documentation: sklearn.gaussian_process.GaussianProcessClassifier

Param Distributions

  1. “base gp classifier”

    n_restarts_optimizer: 5

“hgb classifier”#

Base Class Documentation: sklearn.ensemble.gradient_boosting.HistGradientBoostingClassifier

Param Distributions

  1. “default”

    defaults only

  2. “hgb dist1”

    max_iter: Scalar(init=100, lower=3, upper=200).set_mutation(sigma=32.833333333333336).set_bounds(full_range_sampling=False, lower=3, upper=200).set_integer_casting()

  3. “hgb dist2”

    max_iter: Scalar(init=100, lower=3, upper=500).set_mutation(sigma=82.83333333333333).set_bounds(full_range_sampling=False, lower=3, upper=500).set_integer_casting()
min_samples_leaf: Scalar(lower=10, upper=100).set_mutation(sigma=15.0).set_bounds(full_range_sampling=True, lower=10, upper=100).set_integer_casting()
max_leaf_nodes: Scalar(init=20, lower=6, upper=80).set_mutation(sigma=12.333333333333334).set_bounds(full_range_sampling=False, lower=6, upper=80).set_integer_casting()
l2_regularization: TransitionChoice([0, Log(lower=1e-05, upper=1)])

“knn classifier”#

Base Class Documentation: sklearn.neighbors.KNeighborsClassifier

Param Distributions

  1. “base knn”

    n_neighbors: 5

  2. “knn dist”

    weights: TransitionChoice(['uniform', 'distance'])
n_neighbors: Scalar(lower=2, upper=25).set_mutation(sigma=3.8333333333333335).set_bounds(full_range_sampling=True, lower=2, upper=25).set_integer_casting()

“lasso logistic”#

Base Class Documentation: sklearn.linear_model.LogisticRegression

Param Distributions

  1. “base lasso”

    max_iter: 100
multi_class: 'auto'
penalty: 'l1'
class_weight: None
solver: 'liblinear'

  2. “lasso C”

    max_iter: 100
multi_class: 'auto'
penalty: 'l1'
class_weight: TransitionChoice([None, 'balanced'])
solver: 'liblinear'
C: Log(lower=1e-05, upper=1000.0)

  3. “lasso C extra”

    max_iter: Scalar(lower=100, upper=1000).set_mutation(sigma=150.0).set_bounds(full_range_sampling=True, lower=100, upper=1000).set_integer_casting()
multi_class: 'auto'
penalty: 'l1'
class_weight: TransitionChoice([None, 'balanced'])
solver: 'liblinear'
C: Log(lower=1e-05, upper=1000.0)
tol: Log(lower=1e-06, upper=0.01)

“light gbm classifier”#

Base Class Documentation: BPt.extensions.BPtLGBM.BPtLGBMClassifier

Param Distributions

  1. “base lgbm”

    silent: True

  2. “lgbm classifier dist1”

    silent: True
boosting_type: TransitionChoice(['gbdt', 'dart', 'goss'])
n_estimators: Scalar(init=100, lower=3, upper=500).set_mutation(sigma=82.83333333333333).set_bounds(full_range_sampling=False, lower=3, upper=500).set_integer_casting()
num_leaves: Scalar(init=20, lower=6, upper=80).set_mutation(sigma=12.333333333333334).set_bounds(full_range_sampling=False, lower=6, upper=80).set_integer_casting()
min_child_samples: Scalar(lower=10, upper=500).set_mutation(sigma=81.66666666666667).set_bounds(full_range_sampling=True, lower=10, upper=500).set_integer_casting()
min_child_weight: Log(lower=1e-05, upper=10000.0)
subsample: Scalar(lower=0.3, upper=0.95).set_mutation(sigma=0.10833333333333332).set_bounds(full_range_sampling=True, lower=0.3, upper=0.95)
colsample_bytree: Scalar(lower=0.3, upper=0.95).set_mutation(sigma=0.10833333333333332).set_bounds(full_range_sampling=True, lower=0.3, upper=0.95)
reg_alpha: TransitionChoice([0, Log(lower=1e-05, upper=1)])
reg_lambda: TransitionChoice([0, Log(lower=1e-05, upper=1)])
class_weight: TransitionChoice([None, 'balanced'])

  3. “lgbm classifier dist2”

    silent: True
lambda_l2: 0.001
boosting_type: TransitionChoice(['gbdt', 'dart'])
min_child_samples: TransitionChoice([1, 5, 7, 10, 15, 20, 35, 50, 100, 200, 500, 1000])
num_leaves: TransitionChoice([2, 4, 7, 10, 15, 20, 25, 30, 35, 40, 50, 65, 80, 100, 125, 150, 200, 250])
colsample_bytree: TransitionChoice([0.7, 0.9, 1.0])
subsample: Scalar(lower=0.3, upper=1).set_mutation(sigma=0.11666666666666665).set_bounds(full_range_sampling=True, lower=0.3, upper=1)
learning_rate: TransitionChoice([0.01, 0.05, 0.1])
n_estimators: TransitionChoice([5, 20, 35, 50, 75, 100, 150, 200, 350, 500, 750, 1000])
class_weight: TransitionChoice([None, 'balanced'])

  4. “lgbm classifier dist3”

    silent: True
n_estimators: 1000
early_stopping_rounds: 150
eval_split: 0.2
boosting_type: 'gbdt'
learning_rate: Log(init=0.1, lower=0.005, upper=0.2)
colsample_bytree: Scalar(init=1, lower=0.75, upper=1).set_mutation(sigma=0.041666666666666664).set_bounds(full_range_sampling=False, lower=0.75, upper=1)
min_child_samples: Scalar(init=20, lower=2, upper=30).set_mutation(sigma=4.666666666666667).set_bounds(full_range_sampling=False, lower=2, upper=30).set_integer_casting()
num_leaves: Scalar(init=31, lower=16, upper=96).set_mutation(sigma=13.333333333333334).set_bounds(full_range_sampling=False, lower=16, upper=96).set_integer_casting()
class_weight: TransitionChoice([None, 'balanced'])

“linear svm classifier”#

Base Class Documentation: sklearn.svm.LinearSVC

Param Distributions

  1. “base linear svc”

    max_iter: 100

  2. “linear svc dist”

    max_iter: 100
C: Log(lower=1, upper=10000.0)
class_weight: TransitionChoice([None, 'balanced'])

“logistic”#

Base Class Documentation: sklearn.linear_model.LogisticRegression

Param Distributions

  1. “base logistic”

    max_iter: 100
multi_class: 'auto'
penalty: 'none'
class_weight: None
solver: 'lbfgs'

“pa classifier”#

Base Class Documentation: sklearn.linear_model.PassiveAggressiveClassifier

Param Distributions

  1. “default”

    defaults only

“random forest classifier”#

Base Class Documentation: sklearn.ensemble.RandomForestClassifier

Param Distributions

  1. “base rf regressor”

    n_estimators: 100

  2. “rf classifier dist best”

    n_estimators: Scalar(init=100, lower=10, upper=200).set_mutation(sigma=31.666666666666668).set_bounds(full_range_sampling=False, lower=10, upper=200).set_integer_casting()
class_weight: TransitionChoice([None, 'balanced'])

  3. “rf classifier dist”

    n_estimators: Scalar(init=100, lower=3, upper=500).set_mutation(sigma=82.83333333333333).set_bounds(full_range_sampling=False, lower=3, upper=500).set_integer_casting()
max_depth: TransitionChoice([None, Scalar(init=25, lower=2, upper=200).set_mutation(sigma=33.0).set_bounds(full_range_sampling=False, lower=2, upper=200).set_integer_casting()])
max_features: Scalar(lower=0.1, upper=1.0).set_mutation(sigma=0.15).set_bounds(full_range_sampling=True, lower=0.1, upper=1.0)
min_samples_split: Scalar(lower=0.1, upper=1.0).set_mutation(sigma=0.15).set_bounds(full_range_sampling=True, lower=0.1, upper=1.0)
bootstrap: True
class_weight: TransitionChoice([None, 'balanced'])

“ridge logistic”#

Base Class Documentation: sklearn.linear_model.LogisticRegression

Param Distributions

  1. “base ridge”

    max_iter: 100
penalty: 'l2'
solver: 'saga'

  2. “ridge C”

    max_iter: 100
solver: 'saga'
C: Log(lower=1e-05, upper=1000.0)
class_weight: TransitionChoice([None, 'balanced'])

  3. “ridge C extra”

    max_iter: Scalar(lower=100, upper=1000).set_mutation(sigma=150.0).set_bounds(full_range_sampling=True, lower=100, upper=1000).set_integer_casting()
solver: 'saga'
C: Log(lower=1e-05, upper=1000.0)
class_weight: TransitionChoice([None, 'balanced'])
tol: Log(lower=1e-06, upper=0.01)

“sgd classifier”#

Base Class Documentation: sklearn.linear_model.SGDClassifier

Param Distributions

  1. “default”

    defaults only

  2. “sgd elastic classifier”

    loss: 'squared_epsilon_insensitive'
penalty: 'elasticnet'
alpha: Log(lower=1e-05, upper=100000.0)
l1_ratio: Scalar(lower=0.01, upper=1).set_mutation(sigma=0.165).set_bounds(full_range_sampling=True, lower=0.01, upper=1)
class_weight: TransitionChoice([None, 'balanced'])

  3. “sgd classifier big search”

    loss: TransitionChoice(['hinge', 'log', 'modified_huber', 'squared_hinge', 'perceptron'])
penalty: TransitionChoice(['l2', 'l1', 'elasticnet'])
alpha: Log(lower=1e-05, upper=100.0)
l1_ratio: Scalar(lower=0.01, upper=1).set_mutation(sigma=0.165).set_bounds(full_range_sampling=True, lower=0.01, upper=1)
max_iter: 100
learning_rate: TransitionChoice(['optimal', 'invscaling', 'adaptive', 'constant'])
eta0: Log(lower=1e-06, upper=1000.0)
power_t: Scalar(lower=0.1, upper=0.9).set_mutation(sigma=0.13333333333333333).set_bounds(full_range_sampling=True, lower=0.1, upper=0.9)
early_stopping: TransitionChoice([False, True])
validation_fraction: Scalar(lower=0.05, upper=0.5).set_mutation(sigma=0.075).set_bounds(full_range_sampling=True, lower=0.05, upper=0.5)
n_iter_no_change: Scalar(lower=5, upper=30).set_mutation(sigma=4.166666666666667).set_bounds(full_range_sampling=True, lower=5, upper=30).set_integer_casting()
class_weight: TransitionChoice([None, 'balanced'])

“svm classifier”#

Base Class Documentation: sklearn.svm.SVC

Param Distributions

  1. “base svm classifier”

    kernel: 'rbf'
gamma: 'scale'
probability: True

  2. “svm classifier dist”

    kernel: 'rbf'
gamma: Log(lower=1e-06, upper=1)
C: Log(lower=0.0001, upper=10000.0)
probability: True
class_weight: TransitionChoice([None, 'balanced'])

“xgb classifier”#

Base Class Documentation: xgboost.XGBClassifier

Param Distributions

  1. “base xgb classifier”

    verbosity: 0
objective: 'binary:logistic'

  2. “xgb classifier dist1”

    verbosity: 0
objective: 'binary:logistic'
n_estimators: Scalar(init=100, lower=3, upper=500).set_mutation(sigma=82.83333333333333).set_bounds(full_range_sampling=False, lower=3, upper=500).set_integer_casting()
min_child_weight: Log(lower=1e-05, upper=10000.0)
subsample: Scalar(lower=0.3, upper=0.95).set_mutation(sigma=0.10833333333333332).set_bounds(full_range_sampling=True, lower=0.3, upper=0.95)
colsample_bytree: Scalar(lower=0.3, upper=0.95).set_mutation(sigma=0.10833333333333332).set_bounds(full_range_sampling=True, lower=0.3, upper=0.95)
reg_alpha: TransitionChoice([0, Log(lower=1e-05, upper=1)])
reg_lambda: TransitionChoice([0, Log(lower=1e-05, upper=1)])

  3. “xgb classifier dist2”

    verbosity: 0
objective: 'binary:logistic'
max_depth: TransitionChoice([None, Scalar(init=25, lower=2, upper=200).set_mutation(sigma=33.0).set_bounds(full_range_sampling=False, lower=2, upper=200).set_integer_casting()])
learning_rate: Scalar(lower=0.01, upper=0.5).set_mutation(sigma=0.08166666666666667).set_bounds(full_range_sampling=True, lower=0.01, upper=0.5)
n_estimators: Scalar(lower=3, upper=500).set_mutation(sigma=82.83333333333333).set_bounds(full_range_sampling=True, lower=3, upper=500).set_integer_casting()
min_child_weight: TransitionChoice([1, 5, 10, 50])
subsample: Scalar(lower=0.5, upper=1).set_mutation(sigma=0.08333333333333333).set_bounds(full_range_sampling=True, lower=0.5, upper=1)
colsample_bytree: Scalar(lower=0.4, upper=0.95).set_mutation(sigma=0.09166666666666666).set_bounds(full_range_sampling=True, lower=0.4, upper=0.95)

  4. “xgb classifier dist3”

    verbosity: 0
objective: 'binary:logistic'
learning_rare: Scalar(lower=0.005, upper=0.3).set_mutation(sigma=0.049166666666666664).set_bounds(full_range_sampling=True, lower=0.005, upper=0.3)
min_child_weight: Scalar(lower=0.5, upper=10).set_mutation(sigma=1.5833333333333333).set_bounds(full_range_sampling=True, lower=0.5, upper=10)
max_depth: TransitionChoice(array([3, 4, 5, 6, 7, 8, 9]))
subsample: Scalar(lower=0.5, upper=1).set_mutation(sigma=0.08333333333333333).set_bounds(full_range_sampling=True, lower=0.5, upper=1)
colsample_bytree: Scalar(lower=0.5, upper=1).set_mutation(sigma=0.08333333333333333).set_bounds(full_range_sampling=True, lower=0.5, upper=1)
reg_alpha: Log(lower=1e-05, upper=1)

regression#

“ard regressor”#

Base Class Documentation: sklearn.linear_model.ARDRegression

Param Distributions

  1. “default”

    defaults only

“bayesian ridge regressor”#

Base Class Documentation: sklearn.linear_model.BayesianRidge

Param Distributions

  1. “default”

    defaults only

“dt regressor”#

Base Class Documentation: sklearn.tree.DecisionTreeRegressor

Param Distributions

  1. “default”

    defaults only

  2. “dt dist”

    max_depth: Scalar(lower=1, upper=30).set_mutation(sigma=4.833333333333333).set_bounds(full_range_sampling=True, lower=1, upper=30).set_integer_casting()
min_samples_split: Scalar(lower=2, upper=50).set_mutation(sigma=8.0).set_bounds(full_range_sampling=True, lower=2, upper=50).set_integer_casting()

“elastic net regressor”#

Base Class Documentation: sklearn.linear_model.ElasticNet

Param Distributions

  1. “base elastic net”

    max_iter: 100

  2. “elastic regression”

    max_iter: 100
alpha: Log(lower=1e-05, upper=100000.0)
l1_ratio: Scalar(lower=0.01, upper=1).set_mutation(sigma=0.165).set_bounds(full_range_sampling=True, lower=0.01, upper=1)

  3. “elastic regression extra”

    max_iter: Scalar(lower=100, upper=1000).set_mutation(sigma=150.0).set_bounds(full_range_sampling=True, lower=100, upper=1000).set_integer_casting()
alpha: Log(lower=1e-05, upper=100000.0)
l1_ratio: Scalar(lower=0.01, upper=1).set_mutation(sigma=0.165).set_bounds(full_range_sampling=True, lower=0.01, upper=1)
tol: Log(lower=1e-06, upper=0.01)

“et regressor”#

Base Class Documentation: sklearn.ensemble.ExtraTreesRegressor

Param Distributions

  1. “default”

    defaults only

“gb regressor”#

Base Class Documentation: sklearn.ensemble.GradientBoostingRegressor

Param Distributions

  1. “default”

    defaults only

“gp regressor”#

Base Class Documentation: sklearn.gaussian_process.GaussianProcessRegressor

Param Distributions

  1. “base gp regressor”

    n_restarts_optimizer: 5
normalize_y: True

“hgb regressor”#

Base Class Documentation: sklearn.ensemble.gradient_boosting.HistGradientBoostingRegressor

Param Distributions

  1. “default”

    defaults only

  2. “hgb dist1”

    max_iter: Scalar(init=100, lower=3, upper=200).set_mutation(sigma=32.833333333333336).set_bounds(full_range_sampling=False, lower=3, upper=200).set_integer_casting()

  3. “hgb dist2”

    max_iter: Scalar(init=100, lower=3, upper=500).set_mutation(sigma=82.83333333333333).set_bounds(full_range_sampling=False, lower=3, upper=500).set_integer_casting()
min_samples_leaf: Scalar(lower=10, upper=100).set_mutation(sigma=15.0).set_bounds(full_range_sampling=True, lower=10, upper=100).set_integer_casting()
max_leaf_nodes: Scalar(init=20, lower=6, upper=80).set_mutation(sigma=12.333333333333334).set_bounds(full_range_sampling=False, lower=6, upper=80).set_integer_casting()
l2_regularization: TransitionChoice([0, Log(lower=1e-05, upper=1)])

“knn regressor”#

Base Class Documentation: sklearn.neighbors.KNeighborsRegressor

Param Distributions

  1. “base knn regression”

    n_neighbors: 5

  2. “knn dist regression”

    weights: TransitionChoice(['uniform', 'distance'])
n_neighbors: Scalar(lower=2, upper=25).set_mutation(sigma=3.8333333333333335).set_bounds(full_range_sampling=True, lower=2, upper=25).set_integer_casting()

“lasso regressor”#

Base Class Documentation: sklearn.linear_model.Lasso

Param Distributions

  1. “base lasso regressor”

    max_iter: 100

  2. “lasso regressor dist”

    max_iter: 100
alpha: Log(lower=1e-05, upper=100000.0)

“light gbm regressor”#

Base Class Documentation: BPt.extensions.BPtLGBM.BPtLGBMRegressor

Param Distributions

  1. “base lgbm”

    silent: True

  2. “lgbm dist best”

    silent: True
lambda_l2: 0.001
boosting_type: TransitionChoice(['gbdt', 'dart'])
min_child_samples: TransitionChoice([1, 5, 7, 10, 15, 20, 35, 50, 100, 200, 500, 1000])
num_leaves: TransitionChoice([2, 4, 7, 10, 15, 20, 25, 30, 35, 40, 50, 65, 80, 100, 125, 150, 200, 250])
colsample_bytree: TransitionChoice([0.7, 0.9, 1.0])
subsample: Scalar(lower=0.3, upper=1).set_mutation(sigma=0.11666666666666665).set_bounds(full_range_sampling=True, lower=0.3, upper=1)
learning_rate: TransitionChoice([0.01, 0.05, 0.1])
n_estimators: TransitionChoice([5, 20, 35, 50, 75, 100, 150, 200, 350, 500, 750, 1000])

  3. “lgbm dist1”

    silent: True
boosting_type: TransitionChoice(['gbdt', 'dart', 'goss'])
n_estimators: Scalar(init=100, lower=3, upper=500).set_mutation(sigma=82.83333333333333).set_bounds(full_range_sampling=False, lower=3, upper=500).set_integer_casting()
num_leaves: Scalar(init=20, lower=6, upper=80).set_mutation(sigma=12.333333333333334).set_bounds(full_range_sampling=False, lower=6, upper=80).set_integer_casting()
min_child_samples: Scalar(lower=10, upper=500).set_mutation(sigma=81.66666666666667).set_bounds(full_range_sampling=True, lower=10, upper=500).set_integer_casting()
min_child_weight: Log(lower=1e-05, upper=10000.0)
subsample: Scalar(lower=0.3, upper=0.95).set_mutation(sigma=0.10833333333333332).set_bounds(full_range_sampling=True, lower=0.3, upper=0.95)
colsample_bytree: Scalar(lower=0.3, upper=0.95).set_mutation(sigma=0.10833333333333332).set_bounds(full_range_sampling=True, lower=0.3, upper=0.95)
reg_alpha: TransitionChoice([0, Log(lower=1e-05, upper=1)])
reg_lambda: TransitionChoice([0, Log(lower=1e-05, upper=1)])

  4. “lgbm dist3”

    silent: True
n_estimators: 1000
early_stopping_rounds: 150
eval_split: 0.2
boosting_type: 'gbdt'
learning_rate: Log(init=0.1, lower=0.005, upper=0.2)
colsample_bytree: Scalar(init=1, lower=0.75, upper=1).set_mutation(sigma=0.041666666666666664).set_bounds(full_range_sampling=False, lower=0.75, upper=1)
min_child_samples: Scalar(init=20, lower=2, upper=30).set_mutation(sigma=4.666666666666667).set_bounds(full_range_sampling=False, lower=2, upper=30).set_integer_casting()
num_leaves: Scalar(init=31, lower=16, upper=96).set_mutation(sigma=13.333333333333334).set_bounds(full_range_sampling=False, lower=16, upper=96).set_integer_casting()

“linear regressor”#

Base Class Documentation: sklearn.linear_model.LinearRegression

Param Distributions

  1. “base linear”

    fit_intercept: True

“linear svm regressor”#

Base Class Documentation: sklearn.svm.LinearSVR

Param Distributions

  1. “base linear svr”

    loss: 'epsilon_insensitive'
max_iter: 10000.0

  2. “linear svr dist”

    loss: 'epsilon_insensitive'
max_iter: 10000.0
C: Log(lower=1, upper=10000.0)

“random forest regressor”#

Base Class Documentation: sklearn.ensemble.RandomForestRegressor

Param Distributions

  1. “base rf”

    n_estimators: 100

  2. “rf dist best”

    n_estimators: Scalar(init=100, lower=10, upper=200).set_mutation(sigma=31.666666666666668).set_bounds(full_range_sampling=False, lower=10, upper=200).set_integer_casting()

  3. “rf dist”

    n_estimators: Scalar(init=100, lower=3, upper=500).set_mutation(sigma=82.83333333333333).set_bounds(full_range_sampling=False, lower=3, upper=500).set_integer_casting()
max_depth: TransitionChoice([None, Scalar(init=25, lower=2, upper=200).set_mutation(sigma=33.0).set_bounds(full_range_sampling=False, lower=2, upper=200).set_integer_casting()])
max_features: Scalar(lower=0.1, upper=1.0).set_mutation(sigma=0.15).set_bounds(full_range_sampling=True, lower=0.1, upper=1.0)
min_samples_split: Scalar(lower=0.1, upper=1.0).set_mutation(sigma=0.15).set_bounds(full_range_sampling=True, lower=0.1, upper=1.0)
bootstrap: True

“ridge regressor”#

Base Class Documentation: sklearn.linear_model.Ridge

Param Distributions

  1. “base ridge regressor”

    max_iter: 100
solver: 'lsqr'

  2. “ridge regressor best”

    max_iter: 1000
solver: 'lsqr'
alpha: Log(lower=0.001, upper=1000000.0)

  3. “ridge regressor dist”

    max_iter: 100
solver: 'lsqr'
alpha: Log(lower=0.001, upper=100000.0)

“svm regressor”#

Base Class Documentation: sklearn.svm.SVR

Param Distributions

  1. “base svm”

    kernel: 'rbf'
gamma: 'scale'

  2. “svm dist”

    kernel: 'rbf'
gamma: Log(lower=1e-06, upper=1)
C: Log(lower=0.0001, upper=10000.0)

“tweedie regressor”#

Base Class Documentation: sklearn.linear_model.glm.TweedieRegressor

Param Distributions

  1. “default”

    defaults only

“xgb regressor”#

Base Class Documentation: xgboost.XGBRegressor

Param Distributions

  1. “base xgb”

    verbosity: 0
objective: 'reg:squarederror'

  2. “xgb dist1”

    verbosity: 0
objective: 'reg:squarederror'
n_estimators: Scalar(init=100, lower=3, upper=500).set_mutation(sigma=82.83333333333333).set_bounds(full_range_sampling=False, lower=3, upper=500).set_integer_casting()
min_child_weight: Log(lower=1e-05, upper=10000.0)
subsample: Scalar(lower=0.3, upper=0.95).set_mutation(sigma=0.10833333333333332).set_bounds(full_range_sampling=True, lower=0.3, upper=0.95)
colsample_bytree: Scalar(lower=0.3, upper=0.95).set_mutation(sigma=0.10833333333333332).set_bounds(full_range_sampling=True, lower=0.3, upper=0.95)
reg_alpha: TransitionChoice([0, Log(lower=1e-05, upper=1)])
reg_lambda: TransitionChoice([0, Log(lower=1e-05, upper=1)])

  3. “xgb dist2”

    verbosity: 0
objective: 'reg:squarederror'
max_depth: TransitionChoice([None, Scalar(init=25, lower=2, upper=200).set_mutation(sigma=33.0).set_bounds(full_range_sampling=False, lower=2, upper=200).set_integer_casting()])
learning_rate: Scalar(lower=0.01, upper=0.5).set_mutation(sigma=0.08166666666666667).set_bounds(full_range_sampling=True, lower=0.01, upper=0.5)
n_estimators: Scalar(lower=3, upper=500).set_mutation(sigma=82.83333333333333).set_bounds(full_range_sampling=True, lower=3, upper=500).set_integer_casting()
min_child_weight: TransitionChoice([1, 5, 10, 50])
subsample: Scalar(lower=0.5, upper=1).set_mutation(sigma=0.08333333333333333).set_bounds(full_range_sampling=True, lower=0.5, upper=1)
colsample_bytree: Scalar(lower=0.4, upper=0.95).set_mutation(sigma=0.09166666666666666).set_bounds(full_range_sampling=True, lower=0.4, upper=0.95)

  4. “xgb dist3”

    verbosity: 0
objective: 'reg:squarederror'
learning_rare: Scalar(lower=0.005, upper=0.3).set_mutation(sigma=0.049166666666666664).set_bounds(full_range_sampling=True, lower=0.005, upper=0.3)
min_child_weight: Scalar(lower=0.5, upper=10).set_mutation(sigma=1.5833333333333333).set_bounds(full_range_sampling=True, lower=0.5, upper=10)
max_depth: TransitionChoice(array([3, 4, 5, 6, 7, 8, 9]))
subsample: Scalar(lower=0.5, upper=1).set_mutation(sigma=0.08333333333333333).set_bounds(full_range_sampling=True, lower=0.5, upper=1)
colsample_bytree: Scalar(lower=0.5, upper=1).set_mutation(sigma=0.08333333333333333).set_bounds(full_range_sampling=True, lower=0.5, upper=1)
reg_alpha: Log(lower=1e-05, upper=1)

categorical#

“dt classifier”#

Base Class Documentation: sklearn.tree.DecisionTreeClassifier

Param Distributions

  1. “default”

    defaults only

  2. “dt classifier dist”

    max_depth: Scalar(lower=1, upper=30).set_mutation(sigma=4.833333333333333).set_bounds(full_range_sampling=True, lower=1, upper=30).set_integer_casting()
min_samples_split: Scalar(lower=2, upper=50).set_mutation(sigma=8.0).set_bounds(full_range_sampling=True, lower=2, upper=50).set_integer_casting()
class_weight: TransitionChoice([None, 'balanced'])

“elastic net logistic”#

Base Class Documentation: sklearn.linear_model.LogisticRegression

Param Distributions

  1. “base elastic”

    max_iter: 100
multi_class: 'auto'
penalty: 'elasticnet'
class_weight: None
solver: 'saga'
l1_ratio: 0.5

  2. “elastic classifier”

    max_iter: 100
multi_class: 'auto'
penalty: 'elasticnet'
class_weight: TransitionChoice([None, 'balanced'])
solver: 'saga'
l1_ratio: Scalar(lower=0.01, upper=1).set_mutation(sigma=0.165).set_bounds(full_range_sampling=True, lower=0.01, upper=1)
C: Log(lower=1e-05, upper=100000.0)

  3. “elastic clf v2”

    max_iter: 100
multi_class: 'auto'
penalty: 'elasticnet'
class_weight: TransitionChoice([None, 'balanced'])
solver: 'saga'
l1_ratio: Scalar(lower=0.01, upper=1).set_mutation(sigma=0.165).set_bounds(full_range_sampling=True, lower=0.01, upper=1)
C: Log(lower=0.01, upper=100000.0)

  4. “elastic classifier extra”

    max_iter: Scalar(lower=100, upper=1000).set_mutation(sigma=150.0).set_bounds(full_range_sampling=True, lower=100, upper=1000).set_integer_casting()
multi_class: 'auto'
penalty: 'elasticnet'
class_weight: TransitionChoice([None, 'balanced'])
solver: 'saga'
l1_ratio: Scalar(lower=0.01, upper=1).set_mutation(sigma=0.165).set_bounds(full_range_sampling=True, lower=0.01, upper=1)
C: Log(lower=1e-05, upper=100000.0)
tol: Log(lower=1e-06, upper=0.01)

“et classifier”#

Base Class Documentation: sklearn.ensemble.ExtraTreesClassifier

Param Distributions

  1. “default”

    defaults only

“gaussian nb”#

Base Class Documentation: sklearn.naive_bayes.GaussianNB

Param Distributions

  1. “base gnb”

    var_smoothing: 1e-09

“gb classifier”#

Base Class Documentation: sklearn.ensemble.GradientBoostingClassifier

Param Distributions

  1. “default”

    defaults only

“gp classifier”#

Base Class Documentation: sklearn.gaussian_process.GaussianProcessClassifier

Param Distributions

  1. “base gp classifier”

    n_restarts_optimizer: 5

“hgb classifier”#

Base Class Documentation: sklearn.ensemble.gradient_boosting.HistGradientBoostingClassifier

Param Distributions

  1. “default”

    defaults only

  2. “hgb dist1”

    max_iter: Scalar(init=100, lower=3, upper=200).set_mutation(sigma=32.833333333333336).set_bounds(full_range_sampling=False, lower=3, upper=200).set_integer_casting()

  3. “hgb dist2”

    max_iter: Scalar(init=100, lower=3, upper=500).set_mutation(sigma=82.83333333333333).set_bounds(full_range_sampling=False, lower=3, upper=500).set_integer_casting()
min_samples_leaf: Scalar(lower=10, upper=100).set_mutation(sigma=15.0).set_bounds(full_range_sampling=True, lower=10, upper=100).set_integer_casting()
max_leaf_nodes: Scalar(init=20, lower=6, upper=80).set_mutation(sigma=12.333333333333334).set_bounds(full_range_sampling=False, lower=6, upper=80).set_integer_casting()
l2_regularization: TransitionChoice([0, Log(lower=1e-05, upper=1)])

“knn classifier”#

Base Class Documentation: sklearn.neighbors.KNeighborsClassifier

Param Distributions

  1. “base knn”

    n_neighbors: 5

  2. “knn dist”

    weights: TransitionChoice(['uniform', 'distance'])
n_neighbors: Scalar(lower=2, upper=25).set_mutation(sigma=3.8333333333333335).set_bounds(full_range_sampling=True, lower=2, upper=25).set_integer_casting()

“lasso logistic”#

Base Class Documentation: sklearn.linear_model.LogisticRegression

Param Distributions

  1. “base lasso”

    max_iter: 100
multi_class: 'auto'
penalty: 'l1'
class_weight: None
solver: 'liblinear'

  2. “lasso C”

    max_iter: 100
multi_class: 'auto'
penalty: 'l1'
class_weight: TransitionChoice([None, 'balanced'])
solver: 'liblinear'
C: Log(lower=1e-05, upper=1000.0)

  3. “lasso C extra”

    max_iter: Scalar(lower=100, upper=1000).set_mutation(sigma=150.0).set_bounds(full_range_sampling=True, lower=100, upper=1000).set_integer_casting()
multi_class: 'auto'
penalty: 'l1'
class_weight: TransitionChoice([None, 'balanced'])
solver: 'liblinear'
C: Log(lower=1e-05, upper=1000.0)
tol: Log(lower=1e-06, upper=0.01)

“light gbm classifier”#

Base Class Documentation: BPt.extensions.BPtLGBM.BPtLGBMClassifier

Param Distributions

  1. “base lgbm”

    silent: True

  2. “lgbm classifier dist1”

    silent: True
boosting_type: TransitionChoice(['gbdt', 'dart', 'goss'])
n_estimators: Scalar(init=100, lower=3, upper=500).set_mutation(sigma=82.83333333333333).set_bounds(full_range_sampling=False, lower=3, upper=500).set_integer_casting()
num_leaves: Scalar(init=20, lower=6, upper=80).set_mutation(sigma=12.333333333333334).set_bounds(full_range_sampling=False, lower=6, upper=80).set_integer_casting()
min_child_samples: Scalar(lower=10, upper=500).set_mutation(sigma=81.66666666666667).set_bounds(full_range_sampling=True, lower=10, upper=500).set_integer_casting()
min_child_weight: Log(lower=1e-05, upper=10000.0)
subsample: Scalar(lower=0.3, upper=0.95).set_mutation(sigma=0.10833333333333332).set_bounds(full_range_sampling=True, lower=0.3, upper=0.95)
colsample_bytree: Scalar(lower=0.3, upper=0.95).set_mutation(sigma=0.10833333333333332).set_bounds(full_range_sampling=True, lower=0.3, upper=0.95)
reg_alpha: TransitionChoice([0, Log(lower=1e-05, upper=1)])
reg_lambda: TransitionChoice([0, Log(lower=1e-05, upper=1)])
class_weight: TransitionChoice([None, 'balanced'])

  3. “lgbm classifier dist2”

    silent: True
lambda_l2: 0.001
boosting_type: TransitionChoice(['gbdt', 'dart'])
min_child_samples: TransitionChoice([1, 5, 7, 10, 15, 20, 35, 50, 100, 200, 500, 1000])
num_leaves: TransitionChoice([2, 4, 7, 10, 15, 20, 25, 30, 35, 40, 50, 65, 80, 100, 125, 150, 200, 250])
colsample_bytree: TransitionChoice([0.7, 0.9, 1.0])
subsample: Scalar(lower=0.3, upper=1).set_mutation(sigma=0.11666666666666665).set_bounds(full_range_sampling=True, lower=0.3, upper=1)
learning_rate: TransitionChoice([0.01, 0.05, 0.1])
n_estimators: TransitionChoice([5, 20, 35, 50, 75, 100, 150, 200, 350, 500, 750, 1000])
class_weight: TransitionChoice([None, 'balanced'])

  4. “lgbm classifier dist3”

    silent: True
n_estimators: 1000
early_stopping_rounds: 150
eval_split: 0.2
boosting_type: 'gbdt'
learning_rate: Log(init=0.1, lower=0.005, upper=0.2)
colsample_bytree: Scalar(init=1, lower=0.75, upper=1).set_mutation(sigma=0.041666666666666664).set_bounds(full_range_sampling=False, lower=0.75, upper=1)
min_child_samples: Scalar(init=20, lower=2, upper=30).set_mutation(sigma=4.666666666666667).set_bounds(full_range_sampling=False, lower=2, upper=30).set_integer_casting()
num_leaves: Scalar(init=31, lower=16, upper=96).set_mutation(sigma=13.333333333333334).set_bounds(full_range_sampling=False, lower=16, upper=96).set_integer_casting()
class_weight: TransitionChoice([None, 'balanced'])

“linear svm classifier”#

Base Class Documentation: sklearn.svm.LinearSVC

Param Distributions

  1. “base linear svc”

    max_iter: 100

  2. “linear svc dist”

    max_iter: 100
C: Log(lower=1, upper=10000.0)
class_weight: TransitionChoice([None, 'balanced'])

“logistic”#

Base Class Documentation: sklearn.linear_model.LogisticRegression

Param Distributions

  1. “base logistic”

    max_iter: 100
multi_class: 'auto'
penalty: 'none'
class_weight: None
solver: 'lbfgs'

“pa classifier”#

Base Class Documentation: sklearn.linear_model.PassiveAggressiveClassifier

Param Distributions

  1. “default”

    defaults only

“random forest classifier”#

Base Class Documentation: sklearn.ensemble.RandomForestClassifier

Param Distributions

  1. “base rf regressor”

    n_estimators: 100

  2. “rf classifier dist best”

    n_estimators: Scalar(init=100, lower=10, upper=200).set_mutation(sigma=31.666666666666668).set_bounds(full_range_sampling=False, lower=10, upper=200).set_integer_casting()
class_weight: TransitionChoice([None, 'balanced'])

  3. “rf classifier dist”

    n_estimators: Scalar(init=100, lower=3, upper=500).set_mutation(sigma=82.83333333333333).set_bounds(full_range_sampling=False, lower=3, upper=500).set_integer_casting()
max_depth: TransitionChoice([None, Scalar(init=25, lower=2, upper=200).set_mutation(sigma=33.0).set_bounds(full_range_sampling=False, lower=2, upper=200).set_integer_casting()])
max_features: Scalar(lower=0.1, upper=1.0).set_mutation(sigma=0.15).set_bounds(full_range_sampling=True, lower=0.1, upper=1.0)
min_samples_split: Scalar(lower=0.1, upper=1.0).set_mutation(sigma=0.15).set_bounds(full_range_sampling=True, lower=0.1, upper=1.0)
bootstrap: True
class_weight: TransitionChoice([None, 'balanced'])

“ridge logistic”#

Base Class Documentation: sklearn.linear_model.LogisticRegression

Param Distributions

  1. “base ridge”

    max_iter: 100
penalty: 'l2'
solver: 'saga'

  2. “ridge C”

    max_iter: 100
solver: 'saga'
C: Log(lower=1e-05, upper=1000.0)
class_weight: TransitionChoice([None, 'balanced'])

  3. “ridge C extra”

    max_iter: Scalar(lower=100, upper=1000).set_mutation(sigma=150.0).set_bounds(full_range_sampling=True, lower=100, upper=1000).set_integer_casting()
solver: 'saga'
C: Log(lower=1e-05, upper=1000.0)
class_weight: TransitionChoice([None, 'balanced'])
tol: Log(lower=1e-06, upper=0.01)

“sgd classifier”#

Base Class Documentation: sklearn.linear_model.SGDClassifier

Param Distributions

  1. “default”

    defaults only

  2. “sgd elastic classifier”

    loss: 'squared_epsilon_insensitive'
penalty: 'elasticnet'
alpha: Log(lower=1e-05, upper=100000.0)
l1_ratio: Scalar(lower=0.01, upper=1).set_mutation(sigma=0.165).set_bounds(full_range_sampling=True, lower=0.01, upper=1)
class_weight: TransitionChoice([None, 'balanced'])

  3. “sgd classifier big search”

    loss: TransitionChoice(['hinge', 'log', 'modified_huber', 'squared_hinge', 'perceptron'])
penalty: TransitionChoice(['l2', 'l1', 'elasticnet'])
alpha: Log(lower=1e-05, upper=100.0)
l1_ratio: Scalar(lower=0.01, upper=1).set_mutation(sigma=0.165).set_bounds(full_range_sampling=True, lower=0.01, upper=1)
max_iter: 100
learning_rate: TransitionChoice(['optimal', 'invscaling', 'adaptive', 'constant'])
eta0: Log(lower=1e-06, upper=1000.0)
power_t: Scalar(lower=0.1, upper=0.9).set_mutation(sigma=0.13333333333333333).set_bounds(full_range_sampling=True, lower=0.1, upper=0.9)
early_stopping: TransitionChoice([False, True])
validation_fraction: Scalar(lower=0.05, upper=0.5).set_mutation(sigma=0.075).set_bounds(full_range_sampling=True, lower=0.05, upper=0.5)
n_iter_no_change: Scalar(lower=5, upper=30).set_mutation(sigma=4.166666666666667).set_bounds(full_range_sampling=True, lower=5, upper=30).set_integer_casting()
class_weight: TransitionChoice([None, 'balanced'])

“svm classifier”#

Base Class Documentation: sklearn.svm.SVC

Param Distributions

  1. “base svm classifier”

    kernel: 'rbf'
gamma: 'scale'
probability: True

  2. “svm classifier dist”

    kernel: 'rbf'
gamma: Log(lower=1e-06, upper=1)
C: Log(lower=0.0001, upper=10000.0)
probability: True
class_weight: TransitionChoice([None, 'balanced'])

“xgb classifier”#

Base Class Documentation: xgboost.XGBClassifier

Param Distributions

  1. “base xgb classifier”

    verbosity: 0
objective: 'binary:logistic'

  2. “xgb classifier dist1”

    verbosity: 0
objective: 'binary:logistic'
n_estimators: Scalar(init=100, lower=3, upper=500).set_mutation(sigma=82.83333333333333).set_bounds(full_range_sampling=False, lower=3, upper=500).set_integer_casting()
min_child_weight: Log(lower=1e-05, upper=10000.0)
subsample: Scalar(lower=0.3, upper=0.95).set_mutation(sigma=0.10833333333333332).set_bounds(full_range_sampling=True, lower=0.3, upper=0.95)
colsample_bytree: Scalar(lower=0.3, upper=0.95).set_mutation(sigma=0.10833333333333332).set_bounds(full_range_sampling=True, lower=0.3, upper=0.95)
reg_alpha: TransitionChoice([0, Log(lower=1e-05, upper=1)])
reg_lambda: TransitionChoice([0, Log(lower=1e-05, upper=1)])

  3. “xgb classifier dist2”

    verbosity: 0
objective: 'binary:logistic'
max_depth: TransitionChoice([None, Scalar(init=25, lower=2, upper=200).set_mutation(sigma=33.0).set_bounds(full_range_sampling=False, lower=2, upper=200).set_integer_casting()])
learning_rate: Scalar(lower=0.01, upper=0.5).set_mutation(sigma=0.08166666666666667).set_bounds(full_range_sampling=True, lower=0.01, upper=0.5)
n_estimators: Scalar(lower=3, upper=500).set_mutation(sigma=82.83333333333333).set_bounds(full_range_sampling=True, lower=3, upper=500).set_integer_casting()
min_child_weight: TransitionChoice([1, 5, 10, 50])
subsample: Scalar(lower=0.5, upper=1).set_mutation(sigma=0.08333333333333333).set_bounds(full_range_sampling=True, lower=0.5, upper=1)
colsample_bytree: Scalar(lower=0.4, upper=0.95).set_mutation(sigma=0.09166666666666666).set_bounds(full_range_sampling=True, lower=0.4, upper=0.95)

  4. “xgb classifier dist3”

    verbosity: 0
objective: 'binary:logistic'
learning_rare: Scalar(lower=0.005, upper=0.3).set_mutation(sigma=0.049166666666666664).set_bounds(full_range_sampling=True, lower=0.005, upper=0.3)
min_child_weight: Scalar(lower=0.5, upper=10).set_mutation(sigma=1.5833333333333333).set_bounds(full_range_sampling=True, lower=0.5, upper=10)
max_depth: TransitionChoice(array([3, 4, 5, 6, 7, 8, 9]))
subsample: Scalar(lower=0.5, upper=1).set_mutation(sigma=0.08333333333333333).set_bounds(full_range_sampling=True, lower=0.5, upper=1)
colsample_bytree: Scalar(lower=0.5, upper=1).set_mutation(sigma=0.08333333333333333).set_bounds(full_range_sampling=True, lower=0.5, upper=1)
reg_alpha: Log(lower=1e-05, upper=1)