# plt.show() # def draw_dta(timeSeries): # f = plt.figure(facecolor='white') # timeSeries.plot(color='blue') # plt.show() # def draw_acf_pacf(ts,lags=40): # f = plt.figure(facecolor= 'white') # ax1 = f.add_subplot(211) # plot_acf(ts,lags=40,ax=ax1) # ax2 = f.add_subplot(212) # plot_pacf(ts,lags=40,ax=ax2) # plt.show() test_stationarity.testStationarity(dta) test_stationarity.draw_acf_pacf(dta,l=40) dta_log = np.log(dta) test_stationarity.draw_ts(dta_log) test_stationarity.draw_trend(dta_log,12) # #--------------------------handcraft diff------------------------------------# # diff_12 = dta_log.diff(12) # diff_12.dropna(inplace=True) # diff_12_1 = diff_12.diff(1) # diff_12_1.dropna(inplace=True) # test_stationarity.testStationarity(diff_12_1) # test_stationarity.draw_acf_pacf(diff_12_1)
for row in rows:
# if num == 0:
# last_data = float(row[0])
# if float(row[0]) >400 or float(row[0])<40 :
# data.append(last_data)
# else:
data.append(float(row[0]))
last_data = float(row[0])
time.append(row[1])
num += 1
# x = np.linspace(0, 42, len(data))
# plt.plot(x,data)
# plt.show()
# present = pd.read_csv(file_path,sep = ',')
# print present.shape
# print present.columns
# present_day = present.set_index("data")
# present_day['date'].plot()
# plt.legend(loc = 'best')
# present_day.plot()
# present_day.date.plot(color='g')
# plt.legend(loc = 'best')
# present_day[:10].plot(kind = 'bar')
series_data = pd.Series(data)
draw_trend(series_data, 10)
from test_stationarity import draw_ts
draw_ts(series_data)
from test_stationarity import testStationarity
testStationarity(series_data)
from test_stationarity import draw_acf_pacf
# draw_acf_pacf(series_data)
#test_stationarity.draw_trend(ts_log, 12) #print ts_log ''' diff_12 = ts_log.diff(12) diff_12.dropna(inplace=True) diff_12_1 = diff_12.diff(1) diff_12_1.dropna(inplace=True) print test_stationarity.testStationarity(diff_12_1) ''' rol_mean = ts_log.rolling(window=12).mean() rol_mean.dropna(inplace=True) ts_diff_1 = rol_mean.diff(1) ts_diff_1.dropna(inplace=True) print test_stationarity.testStationarity(ts_diff_1) ts_diff_2 = ts_diff_1.diff(1) ts_diff_2.dropna(inplace=True) print test_stationarity.testStationarity(ts_diff_2) #test_stationarity.draw_acf_pacf(ts_diff_2) from statsmodels.tsa.arima_model import ARMA from statsmodels.tsa import arima_model model = ARMA(ts_diff_2, order=(1, 1)) result_arma = model.fit(disp=-1, method='css') predict_ts = result_arma.predict() # 一阶差分还原
# # temp.extend([test_list[-1], test_list[-1], test_list[-1], test_list[-1]])
# # pyplot.title(str(alpha))
# origin=pyplot.plot(xlabel, test_list, color='red',label='origin')
# mean=pyplot.plot(xlabel, rolmean, color='black',label='mean')
# std=pyplot.plot(xlabel, rolstd, color='green',label='std')
# # pyplot.plot(xlabel, ES(test_list, train_year + month_num, alpha)[-12:], color='black')
# # pyplot.savefig('../result/0921' + u'合并抵消' + '/train_%d_times.jpg' % t)
# pyplot.show()
# test_stationarity.draw_trend(oriData.iloc[i, 1:],6)
rol_weighted_mean = pd.ewma(oriData.iloc[i, 1:], span=6)
test_stationarity.draw_trend(oriData.iloc[i, 1:] - rol_weighted_mean, 6)
# test_stationarity.testStationarity(oriData.iloc[i, 1:])
temp = np.log(oriData.iloc[i, 1:]) - np.log(rol_weighted_mean)
# test_stationarity.testStationarity(oriData.iloc[i, 1:]-rol_weighted_mean)
test_stationarity.testStationarity(temp)
###查分
temp_diff = temp - temp.shift()
test_stationarity.draw_trend(temp_diff, 6)
test_stationarity.testStationarity(temp_diff[1:])
###分解
from statsmodels.tsa.seasonal import seasonal_decompose
# decomposition = seasonal_decompose(oriData.iloc[i,1:], model="additive",freq=6)
decomposition = seasonal_decompose(list(oriData.iloc[i, 1:].values),
model="additive",
freq=6)
trend = decomposition.trend
seasonal = decomposition.seasonal
residual = decomposition.resid
test_stationarity.draw_trend(pd.DataFrame(trend), 6)
import numpy as np
import pandas as pd
from datetime import datetime
import matplotlib.pylab as plt
import test_stationarity
from statsmodels.tsa.seasonal import seasonal_decompose
# 读取数据，pd.read_csv默认生成DataFrame对象，需将其转换成Series对象
df = pd.read_csv('AirPassengers.csv', encoding='utf-8', index_col='date')
df.index = pd.to_datetime(df.index) # 将字符串索引转换成时间索引
ts = df['Passengers'] # 生成pd.Series对象
# 查看数据格式
# print ts.head()
# print ts['1949']
ts_log = np.log(ts)
test_stationarity.draw_ts(ts_log)
test_stationarity.draw_trend(ts_log, 12)
diff_12 = ts_log.diff(12)
diff_12.dropna(inplace=True)
diff_12_1 = diff_12.diff(1)
diff_12_1.dropna(inplace=True)
test_stationarity.testStationarity(diff_12_1)
print(test_stationarity.testStationarity(diff_12_1))
decomposition = seasonal_decompose(ts_log, model="additive")
trend = decomposition.trend
seasonal = decomposition.seasonal
residual = decomposition.resid