import os
from util import ensemble_util
model = ensemble_util.XGBoostModel(
model_path=os.path.abspath(__file__),
corr_threshold=0.9,
search=20,
top_n=3,
eval_func=ensemble_util.xgb_sparse_greedy_f2_metric,
xgb_param={
'eta': [0.1],
'silent': True, # option for logging
'objective':
'binary:logistic', # error evaluation for multiclass tasks
'max_depth': range(2,
11), # depth of the trees in the boosting process
'min_child_weight': [1, 2, 3, 4, 5]
},
number_round=1000,
)
model.train_all_label()
# model.build_and_predict_test()
import os
import sys
sys.path.append(os.path.abspath("../"))
sys.path.append(os.path.abspath("../../"))
from util import ensemble_util
model = ensemble_util.XGBoostModel(
model_path=os.path.abspath(__file__),
corr_threshold=0.9,
search=10,
top_n=1,
eval_func=ensemble_util.xgb_sparse_greedy_f2_metric,
# meta_model_dir="E:\\backup\\jdfc",
meta_model_dir="D:\\github\\JDC\\competition",
xgb_param={
'eta': [0.05],
'silent': True, # option for logging
'objective':
'binary:logistic', # error evaluation for multiclass tasks
'max_depth': range(2,
11), # depth of the trees in the boosting process
'min_child_weight': [1, 2, 3, 4, 5],
'nthread': 6
},
number_round=1000,
)
# model.train_all_label()
model.build_cnn_ensemble()
