1+ import numpy
2+ import matplotlib .pyplot as plt
3+ from pandas import read_csv
4+ import math
5+ from keras .models import Sequential
6+ from keras .layers import Dense
7+ from keras .layers import LSTM
8+ from sklearn .preprocessing import MinMaxScaler
9+ from sklearn .metrics import mean_squared_error
10+ import pickle
11+
12+ # load the dataset
13+ # dataframe = read_csv('1.csv', usecols=[1], engine='python', skipfooter=3)
14+ #dataframe = read_csv('123.csv', usecols=[0], engine='python', skipfooter=3)
15+ #dataset = dataframe.values
16+
17+
18+ def myload (filename ):
19+ abs_path = 'C:/Users/wwwa8/Documents/GitHub/Machine-Learning/序列预测/PCA去趋势化/'
20+ f = open (abs_path + filename ,'rb' )
21+ data = pickle .load (f )
22+ f .close ()
23+ return data
24+
25+ rest_x = myload ("dump_rest_x_9-13.txt" )
26+
27+ rest_x = rest_x .reshape (- 1 ,1 )
28+
29+ dataset = rest_x
30+
31+ # 将整型变为float
32+ dataset = dataset .astype ('float32' )
33+ #plt.plot(dataset,'.')
34+ #plt.plot(dataset,'.')
35+ #plt.show()
36+
37+ # X is the number of passengers at a given time (t) and Y is the number of passengers at the next time (t + 1).
38+ # convert an array of values into a dataset matrix
39+
40+ def create_dataset (dataset , look_back = 1 ):
41+ dataX , dataY = [], []
42+ for i in range (len (dataset )- look_back - 1 ):
43+ a = dataset [i :(i + look_back ), 0 ]
44+ dataX .append (a )
45+ dataY .append (dataset [i + look_back , 0 ])
46+ return numpy .array (dataX ), numpy .array (dataY )
47+
48+ numpy .random .seed (7 )
49+
50+ # normalize the dataset
51+ scaler = MinMaxScaler (feature_range = (0 , 1 ))
52+ dataset = scaler .fit_transform (dataset )
53+
54+ # split into train and test sets
55+ train_size = int (len (dataset ) * 0.67 )
56+ test_size = len (dataset ) - train_size
57+ train , test = dataset [0 :train_size ,:], dataset [train_size :len (dataset ),:]
58+
59+ # use this function to prepare the train and test datasets for modeling
60+ look_back = 1
61+ trainX , trainY = create_dataset (train , look_back )
62+ testX , testY = create_dataset (test , look_back )
63+
64+ # 转换成lstm需要的数据格式
65+ # reshape input to be [samples, time steps, features]
66+ trainX = numpy .reshape (trainX , (trainX .shape [0 ], 1 , trainX .shape [1 ]))
67+ testX = numpy .reshape (testX , (testX .shape [0 ], 1 , testX .shape [1 ]))
68+
69+ # create and fit the LSTM network
70+ model = Sequential ()
71+ model .add (LSTM (3 , input_shape = (1 , look_back )) )
72+ model .add (Dense (1 ))
73+ model .compile (loss = 'mean_squared_error' , optimizer = 'adam' )
74+ model .fit (trainX , trainY , epochs = 30 , batch_size = 1 , verbose = 2 )
75+
76+ # make predictions
77+ trainPredict = model .predict (trainX )
78+ testPredict = model .predict (testX )
79+
80+ # 反标准化
81+ trainPredict = scaler .inverse_transform (trainPredict )
82+ trainY = scaler .inverse_transform ([trainY ])
83+ testPredict = scaler .inverse_transform (testPredict )
84+ testY = scaler .inverse_transform ([testY ])
85+
86+ trainScore = math .sqrt (mean_squared_error (trainY [0 ], trainPredict [:,0 ]))
87+ print ('Train Score: %.2f RMSE' % (trainScore ))
88+ testScore = math .sqrt (mean_squared_error (testY [0 ], testPredict [:,0 ]))
89+ print ('Test Score: %.2f RMSE' % (testScore ))
90+
91+ # shift train predictions for plotting
92+ trainPredictPlot = numpy .empty_like (dataset )
93+ trainPredictPlot [:, :] = numpy .nan
94+ trainPredictPlot [look_back :len (trainPredict )+ look_back , :] = trainPredict
95+
96+ # shift test predictions for plotting
97+ testPredictPlot = numpy .empty_like (dataset )
98+ testPredictPlot [:, :] = numpy .nan
99+ testPredictPlot [len (trainPredict )+ (look_back * 2 )+ 1 :len (dataset )- 1 , :] = testPredict
100+
101+ # plot baseline and predictions
102+ plt .plot (scaler .inverse_transform (dataset ))
103+ plt .plot (trainPredictPlot )
104+ plt .plot (testPredictPlot )
105+ plt .show ()
106+
107+
108+
109+ ###-----------------###
110+
111+ # In [5]: testPredict.shape
112+ # Out[5]: (632, 1)
113+
114+ # In [6]: trainPredict.shape
115+ # Out[6]: (1284, 1)
116+
117+ ###-----------------###
118+
119+ # Test Score: 11.96 RMSE
120+
121+ arr = myload ("dump_arr_9-13.txt" )
122+ main_x = myload ("dump_main_x_9-13.txt" )
123+
124+ arr = arr .reshape (- 1 ,1 )
125+ main_x = main_x .reshape (- 1 ,1 )
126+
127+ ###
128+ main_x_train = main_x [:trainX .shape [0 ]]
129+ main_x_test = main_x [train_size :]
130+ main_x_test = main_x_test [:testX .shape [0 ]]
131+
132+
133+ ###
134+ arr_train = arr [:trainX .shape [0 ]]
135+ arr_test = arr [train_size :]
136+ arr_test = arr_test [:testX .shape [0 ]]
137+
138+ ###add
139+ all_train_predict = main_x_train + trainPredict
140+ all_test_predict = main_x_test + testPredict
141+
142+ def cal_mre (pre_y ,train_y ):
143+ diff = abs (pre_y - train_y )
144+ mre_matrix = diff / train_y
145+ return mre_matrix .mean ()
146+
147+ def replace_ele (train_y ):
148+ train_y [train_y < 0.01 ]= 0.1
149+ return train_y
150+
151+ arr_train = replace_ele (arr_train )
152+ arr_test = replace_ele (arr_test )
153+ all_train_predict = replace_ele (all_train_predict )
154+ all_test_predict = replace_ele (all_test_predict )
155+
156+ print ("train mre : " ,cal_mre (all_train_predict ,arr_train ))
157+ print ("test mre : " ,cal_mre (all_test_predict ,arr_test ))
158+
159+ ##plt
160+ plt .plot (arr_train )
161+ plt .plot (all_train_predict )
162+ plt .show ()
163+
164+ plt .plot (arr_test )
165+ plt .plot (all_test_predict )
166+ plt .show ()
0 commit comments