from cryptoml.util.selection_pipeline import Pipeline
from cryptoml.util.import_proxy import SimpleImputer, StandardScaler, MinMaxScaler, XGBClassifier
PARAMETER_GRID = {}
PARAMETERS = {
"colsample_bylevel": 0.8,
"colsample_bynode": 1,
"colsample_bytree": 0.8,
"learning_rate": 0.3,
"max_depth": 6,
"n_estimators": 500,
"num_parallel_tree": 1,
"reg_alpha": 0,
"reg_lambda": 1,
"subsample": 1,
"use_label_encoder": False,
"seed": None,
"random_state": 0,
"objective": "multi:softmax",
"eval_metric": "mlogloss"
}
estimator = Pipeline([
('i', SimpleImputer(strategy="mean")), # Replace nan's with the mean value between previous and next observation
('s', StandardScaler()), # Standardize data so that Mean and StdDev are < 1
('c', XGBClassifier(**PARAMETERS)),
])
from cryptoml.util.selection_pipeline import Pipeline
from sklearn.ensemble import RandomForestClassifier
from cryptoml.util.import_proxy import SimpleImputer, StandardScaler, MinMaxScaler
PARAMETER_GRID = {
'c__n_estimators': [100, 200, 500],
'i__strategy': ['mean'], # 'median', 'most_frequent', 'constant'
'c__criterion': ['gini'], # , 'entropy'],
'c__max_depth': [2, 3, 4],
'c__min_samples_split': [2],
'c__min_samples_leaf': [1, 0.05, 0.2],
'c__max_features': ['auto'], # 'sqrt',
'c__class_weight': [None, 'balanced'], #, 'balanced_subsample'
}
estimator = Pipeline([
(
'i', SimpleImputer()
), # Replace nan's with the median value between previous and next observation
(
's', StandardScaler()
), # Scale data in order to center it and increase robustness against noise and outliers
('c', RandomForestClassifier()),
])
from cryptoml.util.selection_pipeline import Pipeline
from sklearn.svm import SVC
from cryptoml.util.import_proxy import SimpleImputer, MinMaxScaler, StandardScaler
PARAMETER_GRID = {
'c__C': [1, 5, 10],
# Regularization parameter. The strength of the regularization is inversely proportional to C. >0
'c__kernel': ['poly'],
'c__gamma': ['scale', 'auto'],
# Kernel coefficient for ‘rbf’, ‘poly’ and ‘sigmoid’. (default = 'scale')
'c__degree': [2, 3, 4],
# Degree of the polynomial kernel function (‘poly’). Ignored by all other kernels.
# Independent term in kernel function. It is only significant in ‘poly’ and ‘sigmoid’.
'c__class_weight': [None, 'balanced']
}
estimator = Pipeline([
('i', SimpleImputer()), # Replace nan's with the median value between previous and next observation
('s', StandardScaler()), # Scale data in order to center it and increase robustness against noise and outliers
#('n', MinMaxScaler()), # Scale data in order to center it and increase robustness against noise and outliers
('c', SVC(probability=True)),
])