statsmodels.tools.sm_exceptions.X13Error：错误：预测结束日期，[...]，必须在用户定义的回归变量结束日期当天或之前结束，[...]

Question

我使用下面的测试数据和代码来测试来自

exog

的

sm.tsa.x13_arima_analysis

的

statsmodels

参数，但是出现错误，我该如何解决？非常感谢。

import statsmodels.api as sm
import pandas as pd
import numpy as np
import os

datelist = pd.date_range(pd.to_datetime('2012-12-31'), periods=1500).tolist()
date_df = pd.DataFrame(datelist, columns=['dates'])
endog = pd.concat([date_df, pd.DataFrame(np.random.randint(0, 100, size=(1500, 1)), columns=list('a'))], axis=1)
exog = pd.concat([date_df, pd.DataFrame(np.random.randint(0, 100, size=(1500, 1)), columns=list('b'))], axis=1)
endog.index = endog.dates
endog = endog.drop(['dates'], axis=1)
exog.index = exog.dates
exog = exog.drop(['dates'], axis=1)
endog = endog.resample('M').mean()
exog = exog.resample('M').mean()

X13PATH = r'D:\EconomicDataSeasonalAdjustTool\X13\x13as'
# the link to download X13as model: https://www.census.gov/data/software/x13as.html
os.chdir(X13PATH)
os.environ['X13PATH'] = X13PATH
print(os.environ['X13PATH'])

res = sm.tsa.x13_arima_analysis(endog=endog, exog=exog, x12path=X13PATH, print_stdout=True)
print(res.seasadj)
print(res.plot)

出：

Traceback (most recent call last):
  File "C:\Python\Python310\lib\site-packages\IPython\core\interactiveshell.py", line 3553, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "<ipython-input-12-21406d809098>", line 12, in <cell line: 12>
    res = sm.tsa.x13_arima_analysis(endog=endog, exog=exog, x12path=X13PATH, print_stdout=True)
  File "C:\Users\LSTM\AppData\Roaming\Python\Python310\site-packages\pandas\util\_decorators.py", line 210, in wrapper
    return func(*args, **kwargs)
  File "C:\Users\LSTM\AppData\Local\Programs\Python\Python310\lib\site-packages\statsmodels\tsa\x13.py", line 518, in x13_arima_analysis
    _check_errors(errors)
  File "C:\Users\LSTM\AppData\Local\Programs\Python\Python310\lib\site-packages\statsmodels\tsa\x13.py", line 201, in _check_errors
    raise X13Error(errors)
statsmodels.tools.sm_exceptions.X13Error: ERROR: forecasts end date, 2018.Feb, must end on or before 
        user-defined regression variables end date, 2017.Feb.

Answer 1

最后，下面的代码通过添加

forecast_years=0

:

即可运行

import statsmodels.api as sm
import pandas as pd
import numpy as np
import os
import matplotlib.pyplot as plt

datelist = pd.date_range(pd.to_datetime('2012-12-31'), periods=1500).tolist()
date_df = pd.DataFrame(datelist, columns=['dates'])
endog = pd.concat([date_df, pd.DataFrame(np.random.randint(0, 100, size=(1500, 1)), columns=list('a'))], axis=1)
exog = pd.concat([date_df, pd.DataFrame(np.random.randint(0, 100, size=(1500, 1)), columns=list('b'))], axis=1)
endog.index = endog.dates
endog = endog.drop(['dates'], axis=1)
exog.index = exog.dates
exog = exog.drop(['dates'], axis=1)
endog = endog.resample('M').mean()
exog = exog.resample('M').mean()
exog = exog.reindex(endog.index)

X13PATH = r'D:\zdh_data\a-models\预测实验(3)\X13\x13as'
os.chdir(X13PATH)
os.environ['X13PATH'] = X13PATH
print(os.environ['X13PATH'])

# res = sm.tsa.x13_arima_analysis(endog=endog, exog=exog, x12path=X13PATH, print_stdout=True)
res1 = sm.tsa.x13_arima_analysis(endog=endog, exog=exog, x12path=X13PATH, forecast_years=0, print_stdout=True)
res2 = sm.tsa.x13_arima_analysis(endog=endog, x12path=X13PATH, forecast_years=0, print_stdout=True)
print(res1.seasadj)
print(res2.seasadj)

出：

dates
2012-12-31    48.846047
2013-01-31    62.326384
2013-02-28    54.322351
2013-03-31    52.690840
2013-04-30    47.551649
2013-05-31    54.331202
2013-06-30    50.865211
2013-07-31    48.105752
2013-08-31    48.062653
2013-09-30    48.383561
2013-10-31    57.290482
2013-11-30    42.396296
2013-12-31    50.871920
2014-01-31    46.988795
2014-02-28    54.463265
2014-03-31    46.476195
2014-04-30    50.067421
2014-05-31    51.495410
2014-06-30    55.851400
2014-07-31    45.675858
2014-08-31    51.647678
2014-09-30    39.036736
2014-10-31    65.639687
2014-11-30    49.280582
2014-12-31    56.242091
2015-01-31    44.757873
2015-02-28    42.837859
2015-03-31    54.394856
2015-04-30    50.421756
2015-05-31    52.526607
2015-06-30    49.770681
2015-07-31    57.886844
2015-08-31    35.762811
2015-09-30    51.270080
2015-10-31    45.920629
2015-11-30    54.597159
2015-12-31    42.607645
2016-01-31    52.078086
2016-02-29    56.904964
2016-03-31    49.316221
2016-04-30    54.000531
2016-05-31    43.374732
2016-06-30    45.848618
2016-07-31    49.520627
2016-08-31    49.724006
2016-09-30    50.170454
2016-10-31    41.949666
2016-11-30    52.484160
2016-12-31    50.035846
2017-01-31    47.511667
2017-02-28    41.287805
Freq: M, Name: seasadj, dtype: float64
dates
2012-12-31    49.609184
2013-01-31    62.017442
2013-02-28    54.572534
2013-03-31    52.747710
2013-04-30    47.606578
2013-05-31    54.457986
2013-06-30    50.682544
2013-07-31    48.048137
2013-08-31    48.286644
2013-09-30    47.993489
2013-10-31    56.574971
2013-11-30    42.398613
2013-12-31    51.666707
2014-01-31    46.755879
2014-02-28    54.714097
2014-03-31    46.526357
2014-04-30    50.125257
2014-05-31    51.615576
2014-06-30    55.650827
2014-07-31    45.621153
2014-08-31    51.888377
2014-09-30    38.722020
2014-10-31    64.819902
2014-11-30    49.283275
2014-12-31    57.120778
2015-01-31    44.536015
2015-02-28    43.035150
2015-03-31    54.453565
2015-04-30    50.480001
2015-05-31    52.649180
2015-06-30    49.591945
2015-07-31    57.817515
2015-08-31    35.929480
2015-09-30    50.856737
2015-10-31    45.347119
2015-11-30    54.600143
2015-12-31    43.273317
2016-01-31    51.819943
2016-02-29    57.167042
2016-03-31    49.369449
2016-04-30    54.062910
2016-05-31    43.475949
2016-06-30    45.683967
2016-07-31    49.461317
2016-08-31    49.955740
2016-09-30    49.765976
2016-10-31    41.425749
2016-11-30    52.487029
2016-12-31    50.817571
2017-01-31    47.276159
2017-02-28    41.477957
Freq: M, Name: seasadj, dtype: float64