-
Notifications
You must be signed in to change notification settings - Fork 682
/
Time Series ARIMA SVM
212 lines (167 loc) · 6.11 KB
/
Time Series ARIMA SVM
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
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
import pandas as pd
import matplotlib.pyplot as plt
import numpy
import numpy as np
import pandas
import math
from sklearn.preprocessing import MinMaxScaler
from sklearn.metrics import mean_squared_error
from scipy.interpolate import spline
from sklearn.svm import SVR
from pandas.tools.plotting import autocorrelation_plot
from statsmodels.tsa.arima_model import ARIMA
from scipy.stats import gaussian_kde
from statsmodels.tsa.stattools import adfuller
from statsmodels.tsa.seasonal import seasonal_decompose
def norm(x):
return (x-np.min(x))/(np.max(x)-np.min(x))
dataframe = pd.read_csv('Apple_Data_300.csv').ix[2000:2555,:]
dataframe.head()
autocorrelation_plot(dataframe.ix[:,4])
look_back=50
### AVALIAR V3 LINHAS
model00 = ARIMA(np.array(dataframe.ix[:,4]), dates=None,order=(5,2,2))
model11 = model00.fit(disp=1)
model11.summary()
model11.forecast()
resid9=model11.resid
np.mean(abs(resid9))/max(np.array(dataframe.ix[:,4]))
x3 = resid9
x3 = x3[numpy.logical_not(numpy.isnan(x3))]
dftest13 = adfuller(x3, autolag='AIC')
dfoutput1 = pd.Series(dftest13[0:4], index=['Test Statistic','p-value','#Lags Used','Number of Observations Used'])
print('Dickey Fuller Test:\n',dfoutput1)
start=0
end=len(resid9)
lag=look_back
xx=np.array(resid9[start+lag:end])
yy=np.array(resid9[start:end-lag])
autocorrelation=np.corrcoef(xx,yy)
print('Autocorrelation of Residuals=',round(autocorrelation[0][1],3))
plt.plot(resid9)
plt.title('Residuals ARIMA')
plt.ylim(-50,50)
plt.show()
### FIX
plt.plot(resid9/np.array(dataframe.ix[2002:2555,4]))
plt.title('Residuals/Stock Value - ARIMA')
plt.ylim(-.2,.2)
plt.show()
print(pd.DataFrame(resid9).describe())
plt.hist(resid9)
density = gaussian_kde(resid9)
xs = np.linspace(-50,50,len(resid9))
density.covariance_factor = lambda : .25
density._compute_covariance()
plt.plot(xs,density(xs))
plt.show()
### DELETE OUTLIERS
delete=np.concatenate([np.where(resid9<np.mean(resid9)-2*np.std(resid9))[0],np.where(resid9>np.mean(resid9)+2*np.std(resid9))[0]])
train0=np.delete(np.array(dataframe.ix[:,4]),delete)
train=np.sqrt(train0)
rollmean = pd.rolling_mean(train, window=20)
rollstd = pd.rolling_std(train, window=20)
ts_log0 = np.log(train)
ts_log=pd.DataFrame(ts_log0).dropna()
decomposition = seasonal_decompose(np.array(ts_log).reshape(len(ts_log),),freq=100)
trend = decomposition.trend
seasonal = decomposition.seasonal
residual = decomposition.resid
z=np.where(seasonal==min(seasonal))[0]
period=z[2]-z[1]
look_back = period
plt.figure(figsize=(8,8))
plt.subplot(411)
plt.plot(ts_log, label='Original')
plt.legend(loc='upper left')
plt.subplot(412)
plt.plot(trend, label='Trend',color='red')
plt.legend(loc='upper left')
plt.subplot(413)
plt.plot(seasonal,label='Seasonality',color='green')
plt.legend(loc='upper left')
plt.subplot(414)
plt.plot(residual, label='Residuals',color='black')
plt.legend(loc='upper left')
plt.tight_layout()
from statsmodels.tsa.stattools import adfuller
dftest = adfuller(train, autolag='AIC')
dfoutput = pd.Series(dftest[0:4], index=['Test Statistic','p-value','#Lags Used','Number of Observations Used'])
dfoutput
'''Not Stationary'''
x = seasonal
x = x[numpy.logical_not(numpy.isnan(x))]
dftest1 = adfuller(x, autolag='AIC')
dfoutput1 = pd.Series(dftest1[0:4], index=['Test Statistic','p-value','#Lags Used','Number of Observations Used'])
print('Dickey Fuller Test:\n',dfoutput1)
train=np.sqrt(train0)
for i in range(0,3):
modelP2= ARIMA(np.array(train)[0:-2], order=(2,1,0))
model_fit2 = modelP2.fit(disp=-1,tol=1e-20,maxiter=20000)
pred71 = model_fit2.forecast()[0]
new=np.concatenate((train,pred71),axis=0)
train=new
model_fit2.summary()
print('Precision=',round(float((pred71[-1]**2)/train0[-1]),3))
print('Error=',round(100*(1-float((pred71[-1]**2)/train0[-1])),3),'percent')
print('Real Stock Value',train0[-1])
print('Predicted Stock Value',pred71[-1]**2)
predicted=train**2
predicted_ok=predicted[-4:]
dataframe3 = pd.read_csv('Apple_Data_Comparison.csv')
real_data=np.array(dataframe3.ix[2554:2557,4])
plt.plot(predicted_ok,marker='o',linewidth=2,color='red')
plt.plot(real_data,marker='o',linewidth=2,color='blue')
plt.ylim(130,150)
plt.title('ARIMA PREDICTION')
plt.ylabel('Stock Value')
plt.xlabel('Future Predictions')
plt.show()
predicted_ok-real_data
plt.figure(figsize=(10,6))
line1,=plt.plot(train,color='blue',label='Time Series AAPL')
line2,=plt.plot(rollmean,color='red',label='Rolling Mean',linewidth=2)
line3,=plt.plot(rollstd,color='green',label='Standand Deviation',linewidth=2)
plt.legend([line1,line2,line3],loc='upper left')
plt.show()
pred = []
for i in range(period,len(train)-1):
modelP= ARIMA(np.array(train)[0:i], order=(2,1,0))
model_fit = modelP.fit(disp=0,tol=1e-20,transparams=True,trend='c')
pred7 = model_fit.forecast()[0]
pred.append(pred7)
print('Error=',1-float((pred[-1]**2)/train0[-1]))
plt.plot(np.array(train[period:]),color='blue')
plt.plot(np.array(pred).reshape(len(pred),),color='red')
############### SVMs
dataframe2 = pd.read_csv('Apple_Data_300_SVM.csv')[2000:2555]
look_back=20
train=np.sqrt(np.delete(np.array(dataframe.ix[2000:2555,4]),delete))
dataset0 = dataframe2.values
dataset1 = dataset0.astype('float32')
numpy.random.seed(7)
scaler = MinMaxScaler(feature_range=(0, 1))
dataset = scaler.fit_transform(train)
train_size = int(len(dataset) * .99)
test_size = len(dataset) - train_size
train, test = dataset[0:train_size], dataset[train_size:len(dataset)]
print(len(train), len(test))
def create_dataset(dataset, look_back=1):
dataX, dataY = [], []
for i in range(len(dataset)-look_back):
a = dataset[i:(i+look_back)]
dataX.append(a)
dataY.append(dataset[i + look_back])
return numpy.array(dataX), numpy.array(dataY)
trainX, trainY = create_dataset(train, look_back)
# reshape input to be [samples, time steps, features]
trainY = trainY.reshape(len(trainY), 1)
svr_rbf = SVR(kernel='linear', C=1e3, gamma=0.002)
model = svr_rbf.fit(trainX,trainY.ravel())
model.get_params()
trainPredict = model.predict(trainX)
plt.plot(trainPredict,linewidth=2,color='red')
plt.plot(trainY,linewidth=2,color='blue')
plt.show()
print('Accuracy Train:',1-np.mean(abs(trainPredict-trainY)))
print('Difference Last:',float(trainPredict[-1]-trainY[-1]))