pyhon 進階分析 - 2.2 LightGBM 分類 | 信用卡違約預測 ( Credit Card Default Prediction in python)

[@play_data]

# 延續 3.1 程式碼
params = {
'boosting_type': 'dart', #生成方式 gbdt, dart, rf
'n_estimators' : 1000,
'learning_rate': 0.05,
'n_jobs' : -1, #執行所有CPU
'random_state' : 7,
}
scoring = {
'accuracy' : make_scorer(accuracy_score),
'precision' : make_scorer(precision_score),
'recall' : make_scorer(recall_score),
'f1_score' : make_scorer(f1_score),
}
param_grid = {
'learning_rate': [0.1, 0.05, 0.01],
'max_depth': [3, 5, 7],
'num_leaves': [15, 31, 63],
'n_estimators': [1000, 1500, 2000],
'boosting_type' : ['gbdt','dart'],
'random_state' : [7],
}
train_X_fs, feature_name = permutation_selection(train_X, train_Y,
params = params,
imp = 0.005,
Y_type = 1)
model, cv = build_model(train_X_fs, train_Y, params = params,
Y_type = 1, scoring = scoring)
cv['test_accuracy'].mean()
cv['test_precision'].mean()
cv['test_recall'].mean()
cv['test_f1_score'].mean()
model_grid_tune = grid_tune(train_X_fs, train_Y, Y_type = 1)
model_random_tune = random_tune(train_X_fs, train_Y, Y_type = 1)