PED_pred/main.py at master · gu-yaowen/PED_pred · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
import os
import pandas as pd
import numpy as np
import warnings
import matplotlib.pyplot as plt
from sklearn.experimental import enable_iterative_imputer
from sklearn.impute import IterativeImputer
import xgboost
from sklearn.neighbors import KNeighborsRegressor
from sklearn.ensemble import RandomForestRegressor
from sklearn.model_selection import KFold
from modeling import grid_search, metirc_cal, KFold_train, xgb_model, \
    knn_model, lr_model, lasso_model, rf_model, xgb_model_2, \
    knn_model_2, rf_model_2, lgb_model, lgb_model_2, lgb_clf
from lightgbm import LGBMRegressor
from lightgbm import LGBMClassifier
import joblib
from pandas.core.common import SettingWithCopyWarning

warnings.simplefilter(action="ignore", category=SettingWithCopyWarning)
plt.rcParams['font.family'] = ['Arial Unicode MS']

os.chdir('D:\\儿科')
df_feat = pd.read_csv('feature_0704.csv', encoding='gbk')
# df_feat = df_feat[df_feat['triage'] != 2]
# df_feat = df_feat[df_feat['取号-首诊间隔'] <= 480]
df_feat['取号时间'] = pd.to_datetime(df_feat['取号时间'])
# df_feat = df_feat.drop(['treat.triage.green',
#                         'treat.triage.yellow', 'treat.triage.red'], axis=1)
# df_feat['left.wait.green'] += df_feat['left.nowait.green']
# df_feat['left.wait.other'] += df_feat['left.nowait.other']
# df_feat = df_feat.drop(['left.nowait.green', 'left.nowait.other',
#                         'lwbs.green', 'lwbs.other', 'arrivals.all'], axis=1)
print(np.array(df_feat.columns[3:]))
df_feat.index = df_feat['取号时间']
df_feat_val = df_feat.loc['2021-06-16':]
df_feat = df_feat.loc[:'2021-06-15']

imp = IterativeImputer(max_iter=10, random_state=0)
label = df_feat.iloc[:, 2].values
data = imp.fit_transform(df_feat.iloc[:, 3:].values, label)
label_val = df_feat_val.iloc[:, 2].values
data_val = imp.transform(df_feat_val.iloc[:, 3:].values)
# if 'feature_imputed.npy' not in os.listdir():
#     imp = IterativeImputer(max_iter=10, random_state=0)
#     label = df_feat.iloc[:, 2].values
#     data = imp.fit_transform(df_feat.iloc[:, 3:].values, label)
#     label_val = df_feat_val.iloc[:, 2].values
#     data_val = imp.transform(df_feat_val.iloc[:, 3:].values)
#
#     np.save('feature_imputed_label.npy', label)
#     np.save('feature_imputed_data.npy', data)
#     np.save('feature_imputed_label_val.npy', label_val)
#     np.save('feature_imputed_data_val.npy', data_val)
# else:
#     label = np.load('feature_imputed_label.npy')
#     data = np.load('feature_imputed_data.npy')
#     label_val = np.load('feature_imputed_label_val.npy')
#     data_val = np.load('feature_imputed_data_val.npy')

# knn = KNeighborsRegressor()
# params = {'n_neighbors': range(2, 20, 2), 'leaf_size': range(2, 20, 2)}
# print(grid_search(knn, params, data, label))
# rf = RandomForestRegressor(random_state=0, n_jobs=-1)
# params = {'n_estimators': [100, 150, 200, 250, 300],
#           'max_depth': range(1, 10, 1)}
# print(grid_search(rf, params, data, label))
# params = {'max_depth': range(3, 8, 1), 'num_leaves': range(5, 100, 5),
#           'learning_rate': [0.001, 0.003, 0.005, 0.008, 0.01, 0.03, 0.05, 0.08, 0.1]}
# lgb = LGBMRegressor(n_estimators=200)
# print(grid_search(lgb, params, data, label))
# xgb = xgboost.XGBRegressor(random_state=0, n_jobs=-1,
#                            min_child_weight=3,
#                            gamma=0)
# params = {'n_estimators': [100, 150, 200], 'max_depth': range(3, 10, 2),
#           'learning_rate': [0.001, 0.003, 0.005, 0.008, 0.01, 0.03, 0.05, 0.08, 0.1]}
# print(grid_search(xgb, params, data, label))


# # #
# modelname_list = ['LR', 'LASSO', 'KNN', 'RF', 'LGB', 'XGB']
# modelname_list = ['LGB', 'XGB']
modelname_list = ['XGB']
# model_list = [lr_model, lasso_model, knn_model, rf_model, lgb_model, xgb_model]
# model_list = [lgb_model, xgb_model]
model_list = [xgb_model]
for i in range(6):
    model = modelname_list[i]
    model_func = model_list[i]

    r2, mae, mape, rmse = [], [], [], []
    r2_val, mae_val, mape_val, rmse_val = [], [], [], []
    auc, precision, recall = [], [], []
    if model == 'LGB':
        feature_importance = pd.DataFrame(index=list(df_feat.columns[3:]))
    print('Model {} training...'.format(model))

    predict_cv = df_feat[['ID', '取号-首诊间隔']]
    predict_ev = df_feat_val[['ID', '取号-首诊间隔']]
    for seed in range(10):
        kf = KFold(n_splits=10, shuffle=True, random_state=seed)
        predict, model_all, metric = KFold_train(data, label,
                                                 model_func(seed), model,
                                                 seed, 'regression', cross=True)
        print('Fold {}: R2 {:.3f}, MAE {:.3f}, MAPE {:.3f}, '
              'RMSE {:.3f}, Error quantile: {:.3f}'.format(seed, metric[0], metric[1],
                                                           metric[2], metric[3], metric[4]))
        predict_cv['Predict_{}'.format(seed)] = predict

        if model == 'LGB':
            feature = np.zeros(data.shape[1])
            for i in model_all:
                feature += i.feature_importances_
            feature /= 10
            feature_importance['Fold_{:d}'.format(seed)] = feature

        r2.append(metric[0])
        mae.append(metric[1])
        mape.append(metric[2])
        rmse.append(metric[3])
        df_feat['Predict'] = predict

        pred = np.zeros(label_val.shape)
        for md in model_all:
            pred += md.predict(data_val) / 10
        predict_ev['Predict_{}'.format(seed)] = pred
        metric_val = metirc_cal(label_val, pred, 'External Validation', 'regression')
        print('Fold {}: R2 {:.3f}, MAE {:.3f}, MAPE {:.3f}, '
              'RMSE {:.3f}, Error quantile: {:.3f}'.format(seed, metric_val[0], metric_val[1],
                                                           metric_val[2], metric_val[3], metric_val[4]))
        r2_val.append(metric_val[0])
        mae_val.append(metric_val[1])
        mape_val.append(metric_val[2])
        rmse_val.append(metric_val[3])

        for i in range(len(model_all)):
            joblib.dump(model_all[i], 'models/{}/{}_{}.model'.format(model, seed, i))

    # predict_cv.to_csv('.\\result\\0704_{}_predict_cv_nodis.csv'.format(model), encoding='gbk')
    # predict_ev.to_csv('.\\result\\0704_{}_predict_ev_nodis.csv'.format(model), encoding='gbk')
    # if model == 'LGB':
    #     feature_importance['Avg'] = feature_importance.mean(axis=1)
    #     feature_importance.to_csv('.\\result\\0704_feature_importance_nodis.csv')
    #
    # df = pd.DataFrame(np.array([r2, mae, mape, rmse,
    #                             r2_val, mae_val, mape_val, rmse_val]))
    # df['Avg'] = df.mean(axis=1)
    # df['Std'] = df.std(axis=1)
    # df.index = ['R2', 'MAE', 'MAPE', 'RMSE',
    #             'R2_val', 'MAE_val', 'MAPE_val', 'RMSE_val']
    # df.to_csv('.\\result\\0704_{}_nodis.csv'.format(model))