尝试使用指数平滑进行预测时出现收敛警告

Question

我使用

ExponentialSmoothing

(

statsmodels

) 中的

Version: 0.10.1

对某些数据进行拟合和预测。为了在设置配置时方便使用，我编写了一个函数

exp_smoothing_forecast

，它接收

np array

数据、配置列表 (

[trend, damped, seasonal, seasonal_periods, use_boxcox, remove_bias]

) 以及要预测的

periods

数量。

from statsmodels.tsa.holtwinters import ExponentialSmoothing

def exp_smoothing_forecast(data, config, periods):
    '''
    Perform Holt Winter’s Exponential Smoothing forecast for periods of time.
    '''
    t,d,s,p,b,r = config
    # define model
    model = ExponentialSmoothing(np.array(data), trend=t, damped=d, seasonal=s, seasonal_periods=p)

    # fit model
    model_fit = model.fit(use_boxcox=b, remove_bias=r)

    # make one step forecast
    return model_fit.forecast(periods) 

我使用的数据是每周的。经过一些训练后，我找到了最佳配置并尝试在整个数据上使用它们。

data_1 = [21725.64924, 20826.08817, 26329.99691, 14830.32665, 18169.88488, 19982.91422, 17257.78265, 22538.69626, 18838.27302, 20139.47118, 19080.43343, 15828.84575, 13451.70394, 17548.52795, 14860.06954, 13985.35829, 13260.54788, 12322.06192, 10867.44784, 10853.70009, 13341.73247, 13263.60543, 14649.75235, 14326.58656, 22039.07262, 22888.31507, 32067.63148, 21074.63281, 53075.97138, 16345.24709, 17051.97335, 18658.77237, 21884.1406 , 21657.54719, 21223.70341, 19344.71704, 23775.96044, 22271.27948, 21468.25012, 18073.62045, 9644.77351, 12727.3657 , 15356.19687, 15448.17014, 14723.84147, 17009.4649 , 16092.80642, 16419.31814, 16171.54582, 17271.33481, 21320.07636, 21296.02374, 23199.74761, 16639.21095, 21113.53543, 22749.2432 , 19140.13832, 31088.52782, 16078.59607, 18998.75585, 17151.11463, 16145.82339, 16727.5124 , 16003.41356, 14249.88642, 15841.52938, 13694.78793, 14715.99812, 14128.37359, 13356.26061, 12474.46796, 13246.16726, 13539.81145, 12930.23627, 14233.11001, 16607.7259 , 18365.03853, 20228.55651, 19625.64057, 20626.35991, 51911.0478 , 16432.92248, 18310.47551, 20430.6701 , 26044.62524, 22668.80716, 25020.77289, 23960.46101, 26476.84115, 19499.50768, 23096.5695 , 20812.98895,  8688.23701, 11360.55672, 13862.82367, 17024.09926, 14096.92062, 14283.55523, 16212.13182, 14819.90843, 16950.35181, 16083.05382, 16607.61269, 21116.02172, 19595.84582, 21126.61918, 21906.62752, 19216.92105, 16382.74362, 16425.23783, 21322.25379, 25429.74081, 19949.7062 , 18469.8872 , 19349.82817, 15669.9687 , 12194.87496, 32376.36077, 13010.62105, 12507.4141 , 11235.50094, 11312.38597,  9698.98942, 10099.33959, 11131.35398, 12235.7158 , 13335.16512, 11724.92757, 16637.36082, 19430.46408, 22844.0694 , 19945.1951 , 44117.41448, 17506.60708, 17381.30709, 15527.95481, 18537.75849, 19760.73208, 22897.52474, 19696.13375, 18927.47444, 22095.42535, 28570.44237, 16710.312  ,  7929.1563 , 9383.84275, 11429.63802, 16010.66023, 12717.17012, 12528.87624, 13020.62612, 14404.17851, 19988.99529, 12783.10448, 18541.92333, 15705.89736, 10150.17117,  9778.30207, 22733.63259, 19276.8964 , 16456.28818, 22112.49009, 17702.4329 , 19155.04909, 16150.12167]
data_2 = [ 466723,  452418,  617826,  326684,  395736,  432655,  373109, 510443,  447645,  490145,  473106,  395717,  318162,  448422, 358116,  324310,  305897,  281990,  241155,  241016,  293771, 281642,  307895,  302873,  449448,  455749,  638509,  419042, 1083541,  308940,  333450,  358659,  424403,  414559,  401659, 374299,  475147,  424921,  417545,  355535,  197972,  266886, 327671,  340291,  325005,  387430,  353811,  354566,  353035, 372316,  465626,  460496,  499137,  357643,  462463,  546774, 457630,  755817,  375498,  477250,  421423,  392857,  415778, 409323,  355134,  403246,  332000,  371962,  346441,  327155, 313235,  342809,  341676,  301465,  337626,  370560,  394997, 429656,  412259,  406436, 1046463,  316611,  362318,  385388, 498224,  439668,  488849,  453815,  525870,  385304,  460586, 413856,  172720,  230966,  279383,  356115,  283973,  294327, 344245,  312181,  366841,  350318,  353869,  459447,  425272, 474133,  491659,  428531,  359783,  381855,  493805,  573676, 462372,  426205,  450398,  386455,  290877,  829583,  326638, 308343,  262156,  254527,  228484,  234804,  248274,  260784, 279731,  252299,  349873,  379531,  450056,  388944,  869849, 317120,  331701,  294900,  343365,  365744,  421099,  360206, 351678,  418177,  550871,  327995,  157403,  183336,  226609, 333601,  261049,  260766,  270743,  303745,  413120,  267795, 394175,  333527,  223721,  224568,  536792,  473902,  405797, 569279,  450162,  483895,  426864]

fcast_1 = exp_smoothing_forecast(data_1, ['add', False, 'mul', 52, False, False], 52)
fcast_2 = exp_smoothing_forecast(data_2, ['mul', True, 'mul', 52, False, False], 52)

每次预测结果看起来都很好且一致，但当我尝试执行

ConvergenceWarning

时，我得到了

fcast_2

。

D:\Anaconda3\envs\tsa_course\lib\site-packages\statsmodels\tsa\holtwinters.py:712: ConvergenceWarning: Optimization failed to converge. Check mle_retvals.
  ConvergenceWarning)

有人可以告诉我这个警告的原因是什么吗？

Answer 1

这似乎只是一个过时的错误：https://github.com/unit8co/darts/issues/241。您可以使用以下代码忽略该错误：

import warnings
from statsmodels.tools.sm_exceptions import ConvergenceWarning
warnings.simplefilter('ignore', ConvergenceWarning)