最新

mix/mix_predict_190.py

@@ -28,7 +28,7 @@ def _score(fact, line):
         up_right = up_right + 1.04
     elif fact[1] == 1:
         up_right = up_right + 1
-    elif fact[1] == 2:
+    elif fact[2] == 1:
         up_right = up_right + 0.96
         up_error = up_error + 0.3
     else:

mix/mix_predict_200.py

@@ -18,30 +18,31 @@ def read_data(path):
 
 
 def _score(fact, line):
+    with open('dnn_predict_dmi_18d.txt', 'a') as f:
+        f.write(str([line[-2], line[-1]]) + "\n")
+
     up_right = 0
     up_error = 0
 
     if fact[0] == 1:
         up_right = up_right + 1.04
     elif fact[1] == 1:
-        up_right = up_right + 1
-    elif fact[1] == 2:
-        up_right = up_right + 0.96
         up_error = up_error + 0.3
+        up_right = up_right + 0.98
     else:
         up_error = up_error + 1
-        up_right = up_right + 0.90
+        up_right = up_right + 0.94
     return up_right,up_error
 
 
-def predict(file_path='', model_path='15min_dnn_seq.h5', idx=-1, row=18, col=20):
+def predict(file_path='', model_path='15min_dnn_seq.h5', idx=-1, row=18, col=18):
     test_x,test_y,lines=read_data(file_path)
 
-    test_x_a = test_x[:,:row*col]
-    test_x_a = test_x_a.reshape(test_x.shape[0], row, col, 1)
+    test_x_a = test_x[:,:18*20]
+    test_x_a = test_x_a.reshape(test_x.shape[0], 18, 20, 1)
     # test_x_b = test_x[:, 9*26:9*26+9*26]
     # test_x_b = test_x_b.reshape(test_x.shape[0], 9, 26, 1)
-    test_x_c = test_x[:,row*col:]
+    test_x_c = test_x[:,18*20:]
 
     model=load_model(model_path)
     score = model.evaluate([test_x_c, test_x_a,], test_y)
@@ -54,7 +55,7 @@ def predict(file_path='', model_path='15min_dnn_seq.h5', idx=-1, row=18, col=20)
     down_error = 0
     down_right = 0
     i = 0
-    result = model.predict([test_x_c, test_x_a])
+    result = model.predict([test_x_c, test_x_a,])
     win_dnn = []
     for r in result:
         fact = test_y[i]
@@ -63,7 +64,7 @@ def predict(file_path='', model_path='15min_dnn_seq.h5', idx=-1, row=18, col=20)
             if r[0] > 0.5 or r[1] > 0.5:
                 pass
         else:
-            if r[0] > 0.7:
+            if r[0] > 0.6:
                 tmp_right,tmp_error = _score(fact, lines[i])
                 up_right = tmp_right + up_right
                 up_error = tmp_error + up_error
@@ -71,12 +72,12 @@ def predict(file_path='', model_path='15min_dnn_seq.h5', idx=-1, row=18, col=20)
             elif r[2] > 0.6:
                 if fact[0] == 1:
                     down_error = down_error + 1
-                    down_right = down_right + 1.1
+                    down_right = down_right + 1.04
                 elif fact[1] == 1:
-                    down_error = down_error + 0.2
+                    down_error = down_error + 0.3
                     down_right = down_right + 0.99
                 else:
-                    down_right = down_right + 0.88
+                    down_right = down_right + 0.92
                 down_num = down_num + 1
 
         i = i + 1
@@ -90,8 +91,7 @@ def predict(file_path='', model_path='15min_dnn_seq.h5', idx=-1, row=18, col=20)
 
 if __name__ == '__main__':
     # predict(file_path='D:\\data\\quantization\\stock181_18d_test.log', model_path='181_18d_mix_6D_ma5_s_seq.h5')
-    predict(file_path='D:\\data\\quantization\\stock217_18d_train1.log', model_path='217_18d_mix_5D_ma5_s_seq.h5', row=18, col=18)
-    # predict(file_path='D:\\data\\quantization\\stock321_28d_train1.log', model_path='321_28d_mix_6D_ma5_s_seq.h5', row=28, col=20)
+    predict(file_path='D:\\data\\quantization\\stock216_18d_train1.log', model_path='216_18d_mix_5D_ma5_s_seq.h5', row=18, col=18)
     # predict(file_path='D:\\data\\quantization\\stock6_test.log', model_path='15m_dnn_seq.h5')
     # multi_predict(model='15_18d')
     # predict_today(20200229, model='11_18d')

mix/mix_predict_180.py

@@ -18,34 +18,34 @@ def read_data(path):
 
 
 def _score(fact, line):
-    with open('dnn_predict_dmi_18d.txt', 'a') as f:
-        f.write(str([line[-2], line[-1]]) + "\n")
-
     up_right = 0
     up_error = 0
 
     if fact[0] == 1:
-        up_right = up_right + 1.1
+        up_right = up_right + 1.04
     elif fact[1] == 1:
-        up_right = up_right + 0.99
-        up_error = up_error + 0.2
+        up_right = up_right + 1
+        up_error = up_error + 0.1
+    elif fact[2] == 1:
+        up_right = up_right + 0.96
+        up_error = up_error + 0.5
     else:
         up_error = up_error + 1
         up_right = up_right + 0.90
     return up_right,up_error
 
 
-def predict(file_path='', model_path='15min_dnn_seq.h5', idx=-1):
+def predict(file_path='', model_path='15min_dnn_seq.h5', idx=-1, row=18, col=20):
     test_x,test_y,lines=read_data(file_path)
 
-    test_x_a = test_x[:,:18*18]
-    test_x_a = test_x_a.reshape(test_x.shape[0], 18, 18, 1)
-    # test_x_b = test_x[:, 9*26:9*26+9*26]
-    # test_x_b = test_x_b.reshape(test_x.shape[0], 9, 26, 1)
-    test_x_c = test_x[:,18*18:]
+    test_x_a = test_x[:,:row*col]
+    test_x_a = test_x_a.reshape(test_x.shape[0], row, col, 1)
+    # test_x_b = test_x[:, row*col:row*col+18*2]
+    # test_x_b = test_x_b.reshape(test_x.shape[0], 18, 2, 1)
+    test_x_c = test_x[:,row*col:]
 
     model=load_model(model_path)
-    score = model.evaluate([test_x_c, test_x_a,], test_y)
+    score = model.evaluate([test_x_c, test_x_a, ], test_y)
     print('MIX', score)
 
     up_num = 0
@@ -55,7 +55,7 @@ def predict(file_path='', model_path='15min_dnn_seq.h5', idx=-1):
     down_error = 0
     down_right = 0
     i = 0
-    result = model.predict([test_x_c, test_x_a,])
+    result = model.predict([test_x_c, test_x_a, ])
     win_dnn = []
     for r in result:
         fact = test_y[i]
@@ -64,7 +64,7 @@ def predict(file_path='', model_path='15min_dnn_seq.h5', idx=-1):
             if r[0] > 0.5 or r[1] > 0.5:
                 pass
         else:
-            if r[0] > 0.8:
+            if r[0] > 0.7:
                 tmp_right,tmp_error = _score(fact, lines[i])
                 up_right = tmp_right + up_right
                 up_error = tmp_error + up_error
@@ -91,7 +91,8 @@ def predict(file_path='', model_path='15min_dnn_seq.h5', idx=-1):
 
 if __name__ == '__main__':
     # predict(file_path='D:\\data\\quantization\\stock181_18d_test.log', model_path='181_18d_mix_6D_ma5_s_seq.h5')
-    predict(file_path='D:\\data\\quantization\\stock186C_18d_train1.log', model_path='186C_18d_mix_6D_ma5_s_seq.h5')
+    # predict(file_path='D:\\data\\quantization\\stock217_18d_train1.log', model_path='218_18d_mix_5D_ma5_s_seq.h5', row=18, col=18)
+    predict(file_path='D:\\data\\quantization\\stock321_28d_test.log', model_path='321_28d_mix_5D_ma5_s_seq.h5', row=28, col=20)
     # predict(file_path='D:\\data\\quantization\\stock6_test.log', model_path='15m_dnn_seq.h5')
     # multi_predict(model='15_18d')
     # predict_today(20200229, model='11_18d')

mix/mix_predict_400.py

@@ -0,0 +1,96 @@
+# -*- encoding:utf-8 -*-
+import numpy as np
+from keras.models import load_model
+import joblib
+
+
+def read_data(path):
+    lines = []
+    with open(path) as f:
+        for line in f.readlines()[:]:
+            line = eval(line.strip())
+            lines.append(line)
+
+    size = len(lines[0])
+    train_x=[s[:size - 2] for s in lines]
+    train_y=[s[size-1] for s in lines]
+    return np.array(train_x, dtype=np.float32),np.array(train_y, dtype=np.float32),lines
+
+
+def _score(fact, line):
+    up_right = 0
+    up_error = 0
+
+    if fact[0] == 1:
+        up_right = up_right + 1.04
+    elif fact[1] == 1:
+        up_error = up_error + 0.3
+        up_right = up_right + 0.98
+    else:
+        up_error = up_error + 1
+        up_right = up_right + 0.94
+    return up_right,up_error
+
+
+def predict(file_path='', model_path='15min_dnn_seq.h5', idx=-1, row=18, col=20):
+    test_x,test_y,lines=read_data(file_path)
+
+    test_x_a = test_x[:,:row*col]
+    test_x_a = test_x_a.reshape(test_x.shape[0], row, col, 1)
+    # test_x_b = test_x[:, row*col:row*col+18*2]
+    # test_x_b = test_x_b.reshape(test_x.shape[0], 18, 2, 1)
+    test_x_c = test_x[:,row*col:]
+
+    model=load_model(model_path)
+    score = model.evaluate([test_x_c, test_x_a, ], test_y)
+    print('MIX', score)
+
+    up_num = 0
+    up_error = 0
+    up_right = 0
+    down_num = 0
+    down_error = 0
+    down_right = 0
+    i = 0
+    result = model.predict([test_x_c, test_x_a, ])
+    win_dnn = []
+    for r in result:
+        fact = test_y[i]
+
+        if idx in [-2]:
+            if r[0] > 0.5 or r[1] > 0.5:
+                pass
+        else:
+            if r[0] > 0.7:
+                tmp_right,tmp_error = _score(fact, lines[i])
+                up_right = tmp_right + up_right
+                up_error = tmp_error + up_error
+                up_num = up_num + 1
+            elif r[2] > 0.6:
+                if fact[0] == 1:
+                    down_error = down_error + 1
+                    down_right = down_right + 1.04
+                elif fact[1] == 1:
+                    down_error = down_error + 0.3
+                    down_right = down_right + 0.98
+                else:
+                    down_right = down_right + 0.92
+                down_num = down_num + 1
+
+        i = i + 1
+    if up_num == 0:
+        up_num = 1
+    if down_num == 0:
+        down_num = 1
+    print('MIX', up_right, up_num, up_right/up_num, up_error/up_num, down_right/down_num, down_error/down_num)
+    return win_dnn,up_right/up_num,down_right/down_num
+
+
+if __name__ == '__main__':
+    # predict(file_path='D:\\data\\quantization\\stock181_18d_test.log', model_path='181_18d_mix_6D_ma5_s_seq.h5')
+    # predict(file_path='D:\\data\\quantization\\stock217_18d_train1.log', model_path='218_18d_mix_5D_ma5_s_seq.h5', row=18, col=18)
+    # predict(file_path='D:\\data\\quantization\\stock400_18d_train1.log', model_path='400_18d_mix_5D_ma5_s_seq.h5', row=18, col=18)
+    predict(file_path='D:\\data\\quantization\\stock403_30d_train1.log', model_path='403_30d_mix_5D_ma5_s_seq_2.h5', row=30, col=20)
+    # predict(file_path='D:\\data\\quantization\\stock6_test.log', model_path='15m_dnn_seq.h5')
+    # multi_predict(model='15_18d')
+    # predict_today(20200229, model='11_18d')

mix/mix_predict_500.py

@@ -0,0 +1,100 @@
+# -*- encoding:utf-8 -*-
+import numpy as np
+from keras.models import load_model
+import joblib
+
+
+def read_data(path):
+    lines = []
+    with open(path) as f:
+        for line in f.readlines()[:]:
+            line = eval(line.strip())
+            lines.append(line)
+
+    size = len(lines[0])
+    train_x=[s[:size - 2] for s in lines]
+    train_y=[s[size-1] for s in lines]
+    return np.array(train_x, dtype=np.float32),np.array(train_y, dtype=np.float32),lines
+
+
+def _score(fact, line):
+    up_right = 0
+    up_error = 0
+
+    if fact[0] == 1:
+        up_right = up_right + 1.12
+    elif fact[1] == 1:
+        up_right = up_right + 1.04
+    elif fact[2] == 1:
+        up_error = up_error + 0.3
+        up_right = up_right + 0.98
+    else:
+        up_error = up_error + 1
+        up_right = up_right + 0.94
+    return up_right,up_error
+
+
+def predict(file_path='', model_path='15min_dnn_seq.h5', idx=-1, row=18, col=20):
+    test_x,test_y,lines=read_data(file_path)
+
+    test_x_a = test_x[:,:row*col]
+    test_x_a = test_x_a.reshape(test_x.shape[0], row, col, 1)
+    # test_x_b = test_x[:, row*col:row*col+18*2]
+    # test_x_b = test_x_b.reshape(test_x.shape[0], 18, 2, 1)
+    test_x_c = test_x[:,row*col:]
+
+    model=load_model(model_path)
+    score = model.evaluate([test_x_c, test_x_a, ], test_y)
+    print('MIX', score)
+
+    up_num = 0
+    up_error = 0
+    up_right = 0
+    down_num = 0
+    down_error = 0
+    down_right = 0
+    i = 0
+    result = model.predict([test_x_c, test_x_a, ])
+    win_dnn = []
+    for r in result:
+        fact = test_y[i]
+
+        if idx in [-2]:
+            if r[0] > 0.5 or r[1] > 0.5:
+                pass
+        else:
+            if r[0] > 0.6 :
+                tmp_right,tmp_error = _score(fact, lines[i])
+                up_right = tmp_right + up_right
+                up_error = tmp_error + up_error
+                up_num = up_num + 1
+            elif r[2] > 0.6 or r[3] > 0.6:
+                if fact[0] == 1:
+                    down_error = down_error + 1
+                    down_right = down_right + 1.14
+                elif fact[1] == 1:
+                    down_error = down_error + 0.5
+                    down_right = down_right + 1.04
+                elif fact[2] == 1:
+                    down_right = down_right + 0.98
+                else:
+                    down_right = down_right + 0.92
+                down_num = down_num + 1
+
+        i = i + 1
+    if up_num == 0:
+        up_num = 1
+    if down_num == 0:
+        down_num = 1
+    print('MIX', up_right, up_num, up_right/up_num, up_error/up_num, down_right/down_num, down_error/down_num)
+    return win_dnn,up_right/up_num,down_right/down_num
+
+
+if __name__ == '__main__':
+    # predict(file_path='D:\\data\\quantization\\stock181_18d_test.log', model_path='181_18d_mix_6D_ma5_s_seq.h5')
+    # predict(file_path='D:\\data\\quantization\\stock217_18d_train1.log', model_path='218_18d_mix_5D_ma5_s_seq.h5', row=18, col=18)
+    # predict(file_path='D:\\data\\quantization\\stock400_18d_train1.log', model_path='400_18d_mix_5D_ma5_s_seq.h5', row=18, col=18)
+    predict(file_path='D:\\data\\quantization\\stock501_28d_train1.log', model_path='501_28d_mix_5D_ma5_s_seq.h5', row=28, col=19)
+    # predict(file_path='D:\\data\\quantization\\stock6_test.log', model_path='15m_dnn_seq.h5')
+    # multi_predict(model='15_18d')
+    # predict_today(20200229, model='11_18d')

mix/mix_predict_by_day_190.py

@@ -51,14 +51,18 @@ def predict(file_path='', model_path='15min_dnn_seq', rows=18, cols=18):
 
             ratio = 1
             if train_x_c[0][-1] == 1:
-                ratio = 1.2
+                ratio = 2
+            elif train_x_c[0][-2] == 1:
+                ratio = 1.6
+            elif train_x_c[0][-3] == 1:
+                ratio = 1.3
 
             if result[0][0]> 0.5:
                 up_num = up_num + ratio
             elif result[0][1] > 0.5:
-                up_num = up_num + 0.3
+                up_num = up_num + 0.5*ratio
             elif result[0][2] > 0.5:
-                down_num = down_num + 0.3
+                down_num = down_num + 0.5*ratio
             else:
                 down_num = down_num + ratio
 
@@ -82,5 +86,5 @@ if __name__ == '__main__':
     # predict(file_path='D:\\data\\quantization\\stock9_18_2.log', model_path='18d_dnn_seq.h5')
     # predict(file_path='D:\\data\\quantization\\stock16_18d_20200310.log', model_path='16_18d_mix_seq')
     # predict(file_path='D:\\data\\quantization\\stock196_18d_20200326.log', model_path='196_18d_mix_6D_ma5_s_seq')
-    predict(file_path='D:\\data\\quantization\\stock321_28d_5D_20200327.log', model_path='321_28d_mix_6D_ma5_s_seq', rows=28, cols=20)
+    predict(file_path='D:\\data\\quantization\\stock321_28d_5D_20200403.log', model_path='321_28d_mix_5D_ma5_s_seq', rows=28, cols=20)
     # predict(file_path='D:\\data\\quantization\\stock9_18_4.log', model_path='18d_dnn_seq.h5')

mix/mix_predict_everyday_200.py

@@ -5,21 +5,22 @@ import random
 
 
 holder_stock_list = [
-                        # 医疗
-                        '000150.SZ', '300300.SZ', '603990.SH', '300759.SZ', '300347.SZ', '300003.SZ', '300253.SZ',
-                        # 5G
-                        '300698.SZ', '600498.SH', '300310.SZ', '600353.SH', '603912.SH', '603220.SH', '300602.SZ', '600260.SH',
-                        # 车联网
-                        '002369.SZ', '002920.SZ', '300020.SZ', '002373.SZ', '002869.SZ',
-                        # 工业互联网
-                        '002184.SZ', '002364.SZ','300310.SZ', '300670.SZ', '300166.SZ', '002169.SZ', '002380.SZ',
-                        # 特高压
-                        '300341.SZ', '300670.SZ', '300018.SZ', '600268.SH', '002879.SZ',
-                        # 基础建设
-                        '300041.SZ', '603568.SH', '000967.SZ', '603018.SH',
-                        # 华为
-                        '300687.SZ','002316.SZ','300339.SZ','300378.SZ','300020.SZ','300634.SZ','002570.SZ',
-                        '600801.SH', '300113.SZ','002555.SZ', '002174.SZ',
+        # 医疗
+        '000150.SZ', '300300.SZ', '603990.SH', '300759.SZ', '300347.SZ', '300003.SZ', '300253.SZ','002421.SZ','300168.SZ','002432.SZ',
+        # 5G
+        '300698.SZ', '600498.SH', '300310.SZ', '600353.SH', '603912.SH', '603220.SH', '300602.SZ', '600260.SH', '002463.SZ','300738.SZ',
+        # 车联网
+        '002369.SZ', '002920.SZ', '300020.SZ', '002373.SZ', '002869.SZ','300098.SZ','300048.SZ','002401.SZ',
+        # 工业互联网
+        '002184.SZ', '002364.SZ','300310.SZ', '300670.SZ', '300166.SZ', '002169.SZ', '002380.SZ','002421.SZ',
+        # 特高压
+        '300341.SZ', '300670.SZ', '300018.SZ', '600268.SH', '002879.SZ','002028.SZ',
+        # 基础建设
+        '300041.SZ', '603568.SH', '000967.SZ', '603018.SH','002062.SZ',
+        # 华为
+        '300687.SZ','002316.SZ','300339.SZ','300378.SZ','300020.SZ','300634.SZ','002570.SZ',
+        '600801.SH', '300113.SZ','002555.SZ', '002174.SZ','600585.SH','600276.SH','002415.SZ','000651.SZ',
+        '300074.SZ'
 ]
 
 ROE_stock_list =   [                      # ROE
@@ -99,10 +100,10 @@ def predict_today(file, day, model='10_18d', log=True):
 
     for line in lines:
         train_x = np.array([line[:size - 1]])
-        train_x_tmp = train_x[:,:18*20]
-        train_x_a = train_x_tmp.reshape(train_x.shape[0], 18, 20, 1)
+        train_x_tmp = train_x[:,:30*19]
+        train_x_a = train_x_tmp.reshape(train_x.shape[0], 30, 19, 1)
         # train_x_b = train_x_tmp.reshape(train_x.shape[0], 18, 24)
-        train_x_c = train_x[:,18*20:]
+        train_x_c = train_x[:,30*19:]
 
         result = model.predict([train_x_c, train_x_a, ])
         # print(result, line[-1])
@@ -162,15 +163,15 @@ def predict_today(file, day, model='10_18d', log=True):
                 with open('D:\\data\\quantization\\predict\\' + str(day) + '_mix.txt', mode='a', encoding="utf-8") as f:
                     f.write(str(line[-1]) + ' ' + stock['name'] + ' ' + stock['sw_industry'] + ' ' + str(concept_detail_list) + ' buy' + '\n')
 
-        elif result[0][1] > 0.5:
-            if stock['ts_code'] in holder_stock_list or stock['ts_code'] in ROE_stock_list:
-                print(stock['ts_code'], stock['name'], '震荡评级')
-        elif result[0][2] > 0.4:
-            if stock['ts_code'] in holder_stock_list or stock['ts_code'] in ROE_stock_list:
-                print(stock['ts_code'], stock['name'], '赶紧卖出')
-        else:
-            if stock['ts_code'] in holder_stock_list or stock['ts_code'] in ROE_stock_list:
-                print(stock['ts_code'], stock['name'], result[0],)
+        # elif result[0][1] > 0.5:
+        #     if stock['ts_code'] in holder_stock_list:
+        #         print(stock['ts_code'], stock['name'], '震荡评级')
+        # elif result[0][2] > 0.4:
+        #     if stock['ts_code'] in holder_stock_list:
+        #         print(stock['ts_code'], stock['name'], '赶紧卖出')
+        # else:
+        #     if stock['ts_code'] in holder_stock_list or stock['ts_code'] in ROE_stock_list:
+        #         print(stock['ts_code'], stock['name'], result[0],)
 
     # print(gainian_map)
     # print(hangye_map)
@@ -241,6 +242,6 @@ if __name__ == '__main__':
     # predict(file_path='D:\\data\\quantization\\stock6_5_test.log', model_path='5d_dnn_seq.h5')
     # predict(file_path='D:\\data\\quantization\\stock6_test.log', model_path='15m_dnn_seq.h5')
     # multi_predict()
-    predict_today("D:\\data\\quantization\\stock216_18d_20200327.log", 20200327, model='216_18d_mix_6D_ma5_s_seq.h5', log=True)
+    predict_today("D:\\data\\quantization\\stock405_30d_20200403.log", 20200403, model='405_30d_mix_5D_ma5_s_seq.h5', log=True)
     # join_two_day(20200305, 20200305)
     # check_everyday(20200311, 20200312)

mix/mix_predict_everyday_500.py

@@ -0,0 +1,146 @@
+# -*- encoding:utf-8 -*-
+import numpy as np
+from keras.models import load_model
+import random
+
+
+zixuan_stock_list = [
+                        # 医疗
+                        '000150.SZ', '300300.SZ', '603990.SH', '300759.SZ', '300347.SZ', '300003.SZ', '300253.SZ','002421.SZ','300168.SZ','002432.SZ',
+                        # 5G
+                        '300698.SZ', '600498.SH', '300310.SZ', '600353.SH', '603912.SH', '603220.SH', '300602.SZ', '600260.SH', '002463.SZ','300738.SZ',
+                        # 车联网
+                        '002369.SZ', '002920.SZ', '300020.SZ', '002373.SZ', '002869.SZ','300098.SZ','300048.SZ','002401.SZ',
+                        # 工业互联网
+                        '002184.SZ', '002364.SZ','300310.SZ', '300670.SZ', '300166.SZ', '002169.SZ', '002380.SZ','002421.SZ',
+                        # 特高压
+                        '300341.SZ', '300670.SZ', '300018.SZ', '600268.SH', '002879.SZ','002028.SZ',
+                        # 基础建设
+                        '300041.SZ', '603568.SH', '000967.SZ', '603018.SH','002062.SZ',
+                        # 华为
+                        '300687.SZ','002316.SZ','300339.SZ','300378.SZ','300020.SZ','300634.SZ','002570.SZ',
+                        '600801.SH', '300113.SZ','002555.SZ', '002174.SZ','600585.SH','600276.SH','002415.SZ','000651.SZ',
+                        '300074.SZ'
+]
+
+ROE_stock_list =   [                      # ROE
+'002976.SZ', '002847.SZ', '002597.SZ', '300686.SZ', '000708.SZ', '603948.SH', '600507.SH', '300401.SZ', '002714.SZ', '600732.SH', '300033.SZ', '300822.SZ', '300821.SZ',
+'002458.SZ', '000708.SZ', '600732.SH', '603719.SH', '300821.SZ', '300800.SZ', '300816.SZ', '300812.SZ', '603195.SH', '300815.SZ', '603053.SH', '603551.SH', '002975.SZ',
+'603949.SH', '002970.SZ', '300809.SZ', '002968.SZ', '300559.SZ', '002512.SZ', '300783.SZ', '300003.SZ', '603489.SH', '300564.SZ', '600802.SH', '002600.SZ',
+'000933.SZ', '601918.SH', '000651.SZ', '002916.SZ', '000568.SZ', '000717.SZ', '600452.SH', '603589.SH', '600690.SH', '603886.SH', '300117.SZ', '000858.SZ', '002102.SZ',
+'300136.SZ', '600801.SH', '600436.SH', '300401.SZ', '002190.SZ', '300122.SZ', '002299.SZ', '603610.SH', '002963.SZ', '600486.SH', '300601.SZ', '300682.SZ', '300771.SZ',
+'000868.SZ', '002607.SZ', '603068.SH', '603508.SH', '603658.SH', '300571.SZ', '603868.SH', '600768.SH', '300760.SZ', '002901.SZ', '603638.SH', '601100.SH', '002032.SZ',
+'600083.SH', '600507.SH', '603288.SH', '002304.SZ', '000963.SZ', '300572.SZ', '000885.SZ', '600995.SH', '300080.SZ', '601888.SH', '000048.SZ', '000333.SZ', '300529.SZ',
+'000537.SZ', '002869.SZ', '600217.SH', '000526.SZ', '600887.SH', '002161.SZ', '600267.SH', '600668.SH', '600052.SH', '002379.SZ', '603369.SH', '601360.SH', '002833.SZ',
+'002035.SZ', '600031.SH', '600678.SH', '600398.SH', '600587.SH', '600763.SH', '002016.SZ', '603816.SH', '000031.SZ', '002555.SZ', '603983.SH', '002746.SZ', '603899.SH',
+'300595.SZ', '300632.SZ', '600809.SH', '002507.SZ', '300198.SZ', '600779.SH', '603568.SH', '300638.SZ', '002011.SZ', '603517.SH', '000661.SZ', '300630.SZ', '000895.SZ',
+'002841.SZ', '300602.SZ', '300418.SZ', '603737.SH', '002755.SZ', '002803.SZ', '002182.SZ', '600132.SH', '300725.SZ', '600346.SH', '300015.SZ', '300014.SZ', '300628.SZ',
+'000789.SZ', '600368.SH', '300776.SZ', '600570.SH', '000509.SZ', '600338.SH', '300770.SZ', '600309.SH', '000596.SZ', '300702.SZ', '002271.SZ', '300782.SZ', '300577.SZ',
+'603505.SH', '603160.SH', '300761.SZ', '603327.SH', '002458.SZ', '300146.SZ', '002463.SZ', '300417.SZ', '600566.SH', '002372.SZ', '600585.SH', '000848.SZ', '600519.SH',
+'000672.SZ', '300357.SZ', '002234.SZ', '603444.SH', '300236.SZ', '603360.SH', '002677.SZ', '300487.SZ', '600319.SH', '002415.SZ', '000403.SZ', '600340.SH', '601318.SH',
+
+]
+
+holder_stock_list = [
+'600498.SH', '002223.SZ', '300136.SZ', '300559.SZ',
+    '600496.SH', '300682.SZ'
+]
+
+import pymongo
+from util.mongodb import get_mongo_table_instance
+code_table = get_mongo_table_instance('tushare_code')
+k_table = get_mongo_table_instance('stock_day_k')
+stock_concept_table = get_mongo_table_instance('tushare_concept_detail')
+all_concept_code_list = list(get_mongo_table_instance('tushare_concept').find({}))
+
+
+industry = ['家用电器', '元器件', 'IT设备', '汽车服务',
+            '汽车配件', '软件服务',
+            '互联网', '纺织',
+            '塑料', '半导体',]
+
+gainian_map = {}
+hangye_map = {}
+
+Z_list = []  # 自选
+R_list = []  #  ROE
+O_list = []  # 其他
+
+def predict_today(file, day, model='10_18d', log=True):
+    lines = []
+    with open(file) as f:
+        for line in f.readlines()[:]:
+            line = eval(line.strip())
+            lines.append(line)
+
+    size = len(lines[0])
+
+    model=load_model(model)
+
+    for line in lines:
+        train_x = np.array([line[:size - 1]])
+        train_x_tmp = train_x[:,:28*20]
+        train_x_a = train_x_tmp.reshape(train_x.shape[0], 28, 20, 1)
+        # train_x_b = train_x_tmp.reshape(train_x.shape[0], 18, 24)
+        train_x_c = train_x[:,28*20:]
+
+        result = model.predict([train_x_c, train_x_a, ])
+        # print(result, line[-1])
+        stock = code_table.find_one({'ts_code':line[-1][0]})
+
+        if result[0][0] > 0.6:
+            pass
+
+        elif result[0][1] > 0.5:
+            pass
+        elif result[0][2] > 0.5 or result[0][3] > 0.5:
+            if stock['ts_code'] in holder_stock_list or stock['ts_code'] in zixuan_stock_list:
+                print(stock['ts_code'], stock['name'], '赶紧卖出')
+        else:
+            pass
+
+    # print(gainian_map)
+    # print(hangye_map)
+
+
+def _read_pfile_map(path):
+    s_list = []
+    with open(path, encoding='utf-8') as f:
+        for line in f.readlines()[:]:
+            s_list.append(line)
+    return s_list
+
+
+def join_two_day(a, b):
+    a_list = _read_pfile_map('D:\\data\\quantization\\predict\\' + str(a) + '.txt')
+    b_list = _read_pfile_map('D:\\data\\quantization\\predict\\dmi_' + str(b) + '.txt')
+    for a in a_list:
+        for b in b_list:
+            if a[2:11] == b[2:11]:
+                print(a)
+
+
+def check_everyday(day, today):
+    a_list = _read_pfile_map('D:\\data\\quantization\\predict\\' + str(day) + '.txt')
+    x = 0
+    for a in a_list:
+        print(a[:-1])
+        k_day_list = list(k_table.find({'code':a[2:11], 'tradeDate':{'$lte':int(today)}}).sort('tradeDate', pymongo.DESCENDING).limit(5))
+        if k_day_list is not None and len(k_day_list) > 0:
+            k_day = k_day_list[0]
+            k_day_0 = k_day_list[-1]
+            k_day_last = k_day_list[1]
+            if ((k_day_last['close'] - k_day_0['pre_close'])/k_day_0['pre_close']) < 0.2:
+                print(k_day['open'], k_day['close'], 100*(k_day['close'] - k_day_last['close'])/k_day_last['close'])
+                x = x + 100*(k_day['close'] - k_day_last['close'])/k_day_last['close']
+
+    print(x/len(a_list))
+
+
+if __name__ == '__main__':
+    # predict(file_path='D:\\data\\quantization\\stock6_5_test.log', model_path='5d_dnn_seq.h5')
+    # predict(file_path='D:\\data\\quantization\\stock6_test.log', model_path='15m_dnn_seq.h5')
+    # multi_predict()
+    predict_today("D:\\data\\quantization\\stock500_28d_20200403.log", 20200403, model='500_28d_mix_5D_ma5_s_seq.h5', log=True)
+    # join_two_day(20200305, 20200305)
+    # check_everyday(20200311, 20200312)

mix/mix_train_200.py

@@ -18,10 +18,10 @@ from keras.callbacks import EarlyStopping
 
 early_stopping = EarlyStopping(monitor='accuracy', patience=5, verbose=2)
 
-epochs= 60
+epochs= 42
 size = 420000 #18W 60W
 file_path = 'D:\\data\\quantization\\stock217_18d_train2.log'
-model_path = '217_18d_mix_5D_ma5_s_seq.h5'
+model_path = '218_18d_mix_5D_ma5_s_seq.h5'
 file_path1='D:\\data\\quantization\\stock217_18d_test.log'
 '''
 ROC
@@ -44,8 +44,10 @@ beta+2+5                                           52,96,38
 
 15 流通市值>30                                     54,97,33
                                                    
-16 流通市值>30 + 5d                                59,98,26
-17 roc放后面, 18*18                                57,97,29
+16 流通市值>30 + 5d                                59,97,53     
+17 roc放后面, 18*18                                58,97,31
+18 roc放外面，18*18，两个cnn                       59,97,28
+19 窗口数180                                   
 '''
 
 def read_data(path, path1=file_path1):
@@ -85,9 +87,9 @@ train_x,train_y,test_x,test_y=read_data(file_path)
 
 train_x_a = train_x[:,:18*18]
 train_x_a = train_x_a.reshape(train_x.shape[0], 18, 18, 1)
-# train_x_b = train_x[:, 9*26:18*26]
-# train_x_b = train_x_b.reshape(train_x.shape[0], 9, 26, 1)
-train_x_c = train_x[:,18*18:]
+train_x_b = train_x[:, 18*18:18*18 + 2*18]
+train_x_b = train_x_b.reshape(train_x.shape[0], 18, 2, 1)
+train_x_c = train_x[:,18*18 + 2*18:]
 
 
 def create_mlp(dim, regress=False):
@@ -155,11 +157,11 @@ cnn_0 = create_cnn(18, 18, 1, kernel_size=(6, 20), size=96, regress=False, outpu
 # cnn_0 = create_cnn(18, 20, 1, kernel_size=(9, 9), size=90, regress=False, output=96)         # 28 97 53
 # cnn_0 = create_cnn(18, 20, 1, kernel_size=(3, 20), size=90, regress=False, output=96)
 # cnn_1 = create_cnn(18, 20, 1, kernel_size=(18, 10), size=80, regress=False, output=96)
-# cnn_1 = create_cnn(9, 26, 1, kernel_size=(2, 14), size=36, regress=False, output=64)
+cnn_1 = create_cnn(18, 2, 1, kernel_size=(6, 2), size=24, regress=False, output=24)
 
 # create the input to our final set of layers as the *output* of both
 # the MLP and CNN
-combinedInput = concatenate([mlp.output, cnn_0.output, ])
+combinedInput = concatenate([mlp.output, cnn_0.output, cnn_1.output])
 
 # our final FC layer head will have two dense layers, the final one
 # being our regression head
@@ -176,7 +178,7 @@ x = Dense(3, activation="softmax")(x)
 # our final model will accept categorical/numerical data on the MLP
 # input and images on the CNN input, outputting a single value (the
 # predicted price of the house)
-model = Model(inputs=[mlp.input, cnn_0.input, ], outputs=x)
+model = Model(inputs=[mlp.input, cnn_0.input, cnn_1.input], outputs=x)
 
 
 print("Starting training ")
@@ -191,7 +193,7 @@ model.compile(loss="categorical_crossentropy", optimizer=opt, metrics=['accuracy
 # train the model
 print("[INFO] training model...")
 model.fit(
-    [train_x_c, train_x_a, ], train_y,
+    [train_x_c, train_x_a, train_x_b], train_y,
     # validation_data=([testAttrX, testImagesX], testY),
     # epochs=int(3*train_x_a.shape[0]/1300),
     epochs=epochs,
@@ -203,13 +205,13 @@ model.save(model_path)
 
 test_x_a = test_x[:,:18*18]
 test_x_a = test_x_a.reshape(test_x.shape[0], 18, 18, 1)
-# test_x_b = test_x[:, 9*26:9*26+9*26]
-# test_x_b = test_x_b.reshape(test_x.shape[0], 9, 26, 1)
-test_x_c = test_x[:,18*18:]
+test_x_b = test_x[:, 18*18:18*18+18*2]
+test_x_b = test_x_b.reshape(test_x.shape[0], 18, 2, 1)
+test_x_c = test_x[:,18*18 + 18*2:]
 
 # make predictions on the testing data
 print("[INFO] predicting house prices...")
-score  = model.evaluate([test_x_c, test_x_a,], test_y)
+score  = model.evaluate([test_x_c, test_x_a, test_x_b], test_y)
 
 print(score)
 print('Test score:', score[0])

mix/mix_train_300.py

@@ -18,10 +18,10 @@ from keras.callbacks import EarlyStopping
 
 early_stopping = EarlyStopping(monitor='accuracy', patience=5, verbose=2)
 
-epochs= 88
-size = 410000 #18W 60W
+epochs= 108
+size = 580000 #共68W
 file_path = 'D:\\data\\quantization\\stock321_28d_train2.log'
-model_path = '321_28d_mix_6D_ma5_s_seq.h5'
+model_path = '321_28d_mix_5D_ma5_s_seq_2.h5'
 file_path1='D:\\data\\quantization\\stock321_28d_test.log'
 row = 28
 col = 20
@@ -36,7 +36,7 @@ col = 20
 11 DMI     28*20          37,101,16
 12 MACD    28*19           
 28d+ma5+5+流通市值>10
-21 DMI     28*20          43,101,11  非常好
+21 DMI     28*20          43,102,9  非常好 
  
        
 30d+close
@@ -52,12 +52,12 @@ col = 20
 def read_data(path, path1=file_path1):
     lines = []
     with open(path) as f:
-        for x in range(size): #610000
+        for x in range(size): #680000
             line = eval(f.readline().strip())
             lines.append(line)
 
     with open(path1) as f:
-        for x in range(40000):
+        for x in range(60000): #6w
             line = eval(f.readline().strip())
             lines.append(line)
 

mix/mix_train_400.py

@@ -0,0 +1,208 @@
+import keras
+# -*- encoding:utf-8 -*-
+import numpy as np
+from keras.models import Sequential
+# 优化方法选用Adam(其实可选项有很多，如SGD)
+from keras.optimizers import Adam
+import random
+from keras.models import load_model
+from imblearn.over_sampling import RandomOverSampler
+from keras.utils import np_utils
+# 用于模型初始化，Conv2D模型初始化、Activation激活函数，MaxPooling2D是池化层
+# Flatten作用是将多位输入进行一维化
+# Dense是全连接层
+from keras.layers import Conv2D, Activation, MaxPool2D, Flatten, Dense,Dropout,Input,MaxPooling2D,BatchNormalization,concatenate
+from keras import regularizers
+from keras.models import Model
+from keras.callbacks import EarlyStopping
+
+early_stopping = EarlyStopping(monitor='accuracy', patience=5, verbose=2)
+
+epochs= 40
+size = 440000 #共68W
+file_path = 'D:\\data\\quantization\\stock403_30d_train2.log'
+model_path = '403_30d_mix_5D_ma5_s_seq_2.h5'
+file_path1='D:\\data\\quantization\\stock403_30d_test.log'
+row = 30
+col = 20
+'''
+0    roc 涨幅int表示  18*18             59,97,46                                       
+1    dmi              24*20             59,98,41
+2    roc              24*18             56,98,41
+3    dmi              30*20             55,98,40       
+3B   dmi    9
+3A   dmi              30*20             53,97,44
+4    roc              30*18             63,98,40
+5    macd             30*19             64,98,39       !
+5_1   macd   9                          62,98,41
+5_2   macd   12                         58,98,43
+9    rsi              30*17             50,97,43
+
+6    dmi+大盘形态     30*20             52,98,39
+7    roc+大盘形态     30*18             59,98,40
+8    macd+大盘形态    30*19             
+'''
+
+def read_data(path, path1=file_path1):
+    lines = []
+    with open(path) as f:
+        for x in range(size): #680000
+            line = eval(f.readline().strip())
+            lines.append(line)
+
+    with open(path1) as f:
+        for x in range(30000): #6w
+            line = eval(f.readline().strip())
+            lines.append(line)
+
+    random.shuffle(lines)
+    print('读取数据完毕')
+
+    d=int(0.85*len(lines))
+    length = len(lines[0])
+
+    train_x=[s[:length - 2] for s in lines[0:d]]
+    train_y=[s[-1] for s in lines[0:d]]
+    test_x=[s[:length - 2] for s in lines[d:]]
+    test_y=[s[-1] for s in lines[d:]]
+
+    print('转换数据完毕')
+
+    ros = RandomOverSampler(random_state=0)
+    X_resampled, y_resampled = ros.fit_sample(np.array(train_x, dtype=np.float32), np.array(train_y, dtype=np.float32))
+
+    print('数据重采样完毕')
+
+    return X_resampled,y_resampled,np.array(test_x, dtype=np.float32),np.array(test_y, dtype=np.float32)
+
+
+train_x,train_y,test_x,test_y=read_data(file_path)
+
+train_x_a = train_x[:,:row*col]
+train_x_a = train_x_a.reshape(train_x.shape[0], row, col, 1)
+# train_x_b = train_x[:, 9*26:18*26]
+# train_x_b = train_x_b.reshape(train_x.shape[0], 9, 26, 1)
+train_x_c = train_x[:,row*col:]
+
+
+def create_mlp(dim, regress=False):
+    # define our MLP network
+    model = Sequential()
+    model.add(Dense(256, input_dim=dim, activation="relu"))
+    model.add(Dropout(0.2))
+    model.add(Dense(256, activation="relu"))
+    model.add(Dense(256, activation="relu"))
+    model.add(Dense(128, activation="relu"))
+
+    # check to see if the regression node should be added
+    if regress:
+        model.add(Dense(1, activation="linear"))
+
+    # return our model
+    return model
+
+
+def create_cnn(width, height, depth, size=48, kernel_size=(5, 6), regress=False, output=24):
+    # initialize the input shape and channel dimension, assuming
+    # TensorFlow/channels-last ordering
+    inputShape = (width, height, 1)
+    chanDim = -1
+
+    # define the model input
+    inputs = Input(shape=inputShape)
+    # x = inputs
+    # CONV => RELU => BN => POOL
+    x = Conv2D(size, kernel_size, strides=2, padding="same")(inputs)
+    x = Activation("relu")(x)
+    x = BatchNormalization(axis=chanDim)(x)
+
+    # y = Conv2D(24, (2, 8), strides=2, padding="same")(inputs)
+    # y = Activation("relu")(y)
+    # y = BatchNormalization(axis=chanDim)(y)
+
+    # flatten the volume, then FC => RELU => BN => DROPOUT
+    x = Flatten()(x)
+    x = Dense(output)(x)
+    x = Activation("relu")(x)
+    x = BatchNormalization(axis=chanDim)(x)
+    x = Dropout(0.2)(x)
+
+    # apply another FC layer, this one to match the number of nodes
+    # coming out of the MLP
+    x = Dense(output)(x)
+    x = Activation("relu")(x)
+
+    # check to see if the regression node should be added
+    if regress:
+        x = Dense(1, activation="linear")(x)
+
+    # construct the CNN
+    model = Model(inputs, x)
+
+    # return the CNN
+    return model
+
+
+# create the MLP and CNN models
+mlp = create_mlp(train_x_c.shape[1], regress=False)
+# cnn_0 = create_cnn(18, 20, 1, kernel_size=(3, 3), size=90, regress=False, output=96)       # 31 97 46
+cnn_0 = create_cnn(row, col, 1, kernel_size=(6, col), size=96, regress=False, output=96)         # 29 98 47
+# cnn_0 = create_cnn(18, 20, 1, kernel_size=(9, 9), size=90, regress=False, output=96)         # 28 97 53
+# cnn_0 = create_cnn(18, 20, 1, kernel_size=(3, 20), size=90, regress=False, output=96)
+# cnn_1 = create_cnn(18, 20, 1, kernel_size=(18, 10), size=80, regress=False, output=96)
+# cnn_1 = create_cnn(9, 26, 1, kernel_size=(2, 14), size=36, regress=False, output=64)
+
+# create the input to our final set of layers as the *output* of both
+# the MLP and CNN
+combinedInput = concatenate([mlp.output, cnn_0.output, ])
+
+# our final FC layer head will have two dense layers, the final one
+# being our regression head
+x = Dense(1024, activation="relu", kernel_regularizer=regularizers.l1(0.003))(combinedInput)
+x = Dropout(0.2)(x)
+x = Dense(1024, activation="relu")(x)
+x = Dense(1024, activation="relu")(x)
+# 在建设一层
+x = Dense(3, activation="softmax")(x)
+
+# our final model will accept categorical/numerical data on the MLP
+# input and images on the CNN input, outputting a single value (the
+# predicted price of the house)
+model = Model(inputs=[mlp.input, cnn_0.input, ], outputs=x)
+
+
+print("Starting training ")
+# h = model.fit(train_x, train_y, batch_size=4096*2, epochs=500, shuffle=True)
+
+# compile the model using mean absolute percentage error as our loss,
+# implying that we seek to minimize the absolute percentage difference
+# between our price *predictions* and the *actual prices*
+opt = Adam(lr=1e-3, decay=1e-3 / 200)
+model.compile(loss="categorical_crossentropy", optimizer=opt, metrics=['accuracy'])
+
+# train the model
+print("[INFO] training model...")
+model.fit(
+    [train_x_c, train_x_a, ], train_y,
+    # validation_data=([testAttrX, testImagesX], testY),
+    # epochs=int(3*train_x_a.shape[0]/1300),
+    epochs=epochs,
+    batch_size=2048, shuffle=True,
+    callbacks=[early_stopping]
+)
+
+model.save(model_path)
+
+test_x_a = test_x[:,:row*col]
+test_x_a = test_x_a.reshape(test_x.shape[0], row, col, 1)
+# test_x_b = test_x[:, 9*26:9*26+9*26]
+# test_x_b = test_x_b.reshape(test_x.shape[0], 9, 26, 1)
+test_x_c = test_x[:,row*col:]
+
+# make predictions on the testing data
+print("[INFO] predicting house prices...")
+score  = model.evaluate([test_x_c, test_x_a,], test_y)
+
+print(score)
+print('Test score:', score[0])
+print('Test accuracy:', score[1])

mix/mix_train_500.py

@@ -0,0 +1,196 @@
+import keras
+# -*- encoding:utf-8 -*-
+import numpy as np
+from keras.models import Sequential
+# 优化方法选用Adam(其实可选项有很多，如SGD)
+from keras.optimizers import Adam
+import random
+from keras.models import load_model
+from imblearn.over_sampling import RandomOverSampler
+from keras.utils import np_utils
+# 用于模型初始化，Conv2D模型初始化、Activation激活函数，MaxPooling2D是池化层
+# Flatten作用是将多位输入进行一维化
+# Dense是全连接层
+from keras.layers import Conv2D, Activation, MaxPool2D, Flatten, Dense,Dropout,Input,MaxPooling2D,BatchNormalization,concatenate
+from keras import regularizers
+from keras.models import Model
+from keras.callbacks import EarlyStopping
+
+early_stopping = EarlyStopping(monitor='accuracy', patience=5, verbose=2)
+
+epochs= 42
+size = 420000 #共68W
+file_path = 'D:\\data\\quantization\\stock501_28d_train2.log'
+model_path = '501_28d_mix_5D_ma5_s_seq.h5'
+file_path1='D:\\data\\quantization\\stock501_28d_test.log'
+row = 28
+col = 19
+'''
+0    dmi            28*20      38,98,51/5     下跌预判非常准                                    
+1    macd           28*19      41,98,53/8       
+       
+'''
+
+def read_data(path, path1=file_path1):
+    lines = []
+    with open(path) as f:
+        for x in range(size): #680000
+            line = eval(f.readline().strip())
+            lines.append(line)
+
+    with open(path1) as f:
+        for x in range(32000): #6w
+            line = eval(f.readline().strip())
+            lines.append(line)
+
+    random.shuffle(lines)
+    print('读取数据完毕')
+
+    d=int(0.85*len(lines))
+    length = len(lines[0])
+
+    train_x=[s[:length - 2] for s in lines[0:d]]
+    train_y=[s[-1] for s in lines[0:d]]
+    test_x=[s[:length - 2] for s in lines[d:]]
+    test_y=[s[-1] for s in lines[d:]]
+
+    print('转换数据完毕')
+
+    ros = RandomOverSampler(random_state=0)
+    X_resampled, y_resampled = ros.fit_sample(np.array(train_x, dtype=np.float32), np.array(train_y, dtype=np.float32))
+
+    print('数据重采样完毕')
+
+    return X_resampled,y_resampled,np.array(test_x, dtype=np.float32),np.array(test_y, dtype=np.float32)
+
+
+train_x,train_y,test_x,test_y=read_data(file_path)
+
+train_x_a = train_x[:,:row*col]
+train_x_a = train_x_a.reshape(train_x.shape[0], row, col, 1)
+# train_x_b = train_x[:, 9*26:18*26]
+# train_x_b = train_x_b.reshape(train_x.shape[0], 9, 26, 1)
+train_x_c = train_x[:,row*col:]
+
+
+def create_mlp(dim, regress=False):
+    # define our MLP network
+    model = Sequential()
+    model.add(Dense(256, input_dim=dim, activation="relu"))
+    model.add(Dropout(0.2))
+    model.add(Dense(256, activation="relu"))
+    model.add(Dense(256, activation="relu"))
+    model.add(Dense(128, activation="relu"))
+
+    # check to see if the regression node should be added
+    if regress:
+        model.add(Dense(1, activation="linear"))
+
+    # return our model
+    return model
+
+
+def create_cnn(width, height, depth, size=48, kernel_size=(5, 6), regress=False, output=24):
+    # initialize the input shape and channel dimension, assuming
+    # TensorFlow/channels-last ordering
+    inputShape = (width, height, 1)
+    chanDim = -1
+
+    # define the model input
+    inputs = Input(shape=inputShape)
+    # x = inputs
+    # CONV => RELU => BN => POOL
+    x = Conv2D(size, kernel_size, strides=2, padding="same")(inputs)
+    x = Activation("relu")(x)
+    x = BatchNormalization(axis=chanDim)(x)
+
+    # y = Conv2D(24, (2, 8), strides=2, padding="same")(inputs)
+    # y = Activation("relu")(y)
+    # y = BatchNormalization(axis=chanDim)(y)
+
+    # flatten the volume, then FC => RELU => BN => DROPOUT
+    x = Flatten()(x)
+    x = Dense(output)(x)
+    x = Activation("relu")(x)
+    x = BatchNormalization(axis=chanDim)(x)
+    x = Dropout(0.2)(x)
+
+    # apply another FC layer, this one to match the number of nodes
+    # coming out of the MLP
+    x = Dense(output)(x)
+    x = Activation("relu")(x)
+
+    # check to see if the regression node should be added
+    if regress:
+        x = Dense(1, activation="linear")(x)
+
+    # construct the CNN
+    model = Model(inputs, x)
+
+    # return the CNN
+    return model
+
+
+# create the MLP and CNN models
+mlp = create_mlp(train_x_c.shape[1], regress=False)
+# cnn_0 = create_cnn(18, 20, 1, kernel_size=(3, 3), size=90, regress=False, output=96)       # 31 97 46
+cnn_0 = create_cnn(row, col, 1, kernel_size=(6, col), size=96, regress=False, output=96)         # 29 98 47
+# cnn_0 = create_cnn(18, 20, 1, kernel_size=(9, 9), size=90, regress=False, output=96)         # 28 97 53
+# cnn_0 = create_cnn(18, 20, 1, kernel_size=(3, 20), size=90, regress=False, output=96)
+# cnn_1 = create_cnn(18, 20, 1, kernel_size=(18, 10), size=80, regress=False, output=96)
+# cnn_1 = create_cnn(9, 26, 1, kernel_size=(2, 14), size=36, regress=False, output=64)
+
+# create the input to our final set of layers as the *output* of both
+# the MLP and CNN
+combinedInput = concatenate([mlp.output, cnn_0.output, ])
+
+# our final FC layer head will have two dense layers, the final one
+# being our regression head
+x = Dense(1024, activation="relu", kernel_regularizer=regularizers.l1(0.003))(combinedInput)
+x = Dropout(0.2)(x)
+x = Dense(1024, activation="relu")(x)
+x = Dense(1024, activation="relu")(x)
+# 在建设一层
+x = Dense(4, activation="softmax")(x)
+
+# our final model will accept categorical/numerical data on the MLP
+# input and images on the CNN input, outputting a single value (the
+# predicted price of the house)
+model = Model(inputs=[mlp.input, cnn_0.input, ], outputs=x)
+
+
+print("Starting training ")
+# h = model.fit(train_x, train_y, batch_size=4096*2, epochs=500, shuffle=True)
+
+# compile the model using mean absolute percentage error as our loss,
+# implying that we seek to minimize the absolute percentage difference
+# between our price *predictions* and the *actual prices*
+opt = Adam(lr=1e-3, decay=1e-3 / 200)
+model.compile(loss="categorical_crossentropy", optimizer=opt, metrics=['accuracy'])
+
+# train the model
+print("[INFO] training model...")
+model.fit(
+    [train_x_c, train_x_a, ], train_y,
+    # validation_data=([testAttrX, testImagesX], testY),
+    # epochs=int(3*train_x_a.shape[0]/1300),
+    epochs=epochs,
+    batch_size=2048, shuffle=True,
+    callbacks=[early_stopping]
+)
+
+model.save(model_path)
+
+test_x_a = test_x[:,:row*col]
+test_x_a = test_x_a.reshape(test_x.shape[0], row, col, 1)
+# test_x_b = test_x[:, 9*26:9*26+9*26]
+# test_x_b = test_x_b.reshape(test_x.shape[0], 9, 26, 1)
+test_x_c = test_x[:,row*col:]
+
+# make predictions on the testing data
+print("[INFO] predicting house prices...")
+score  = model.evaluate([test_x_c, test_x_a,], test_y)
+
+print(score)
+print('Test score:', score[0])
+print('Test accuracy:', score[1])