做个广告,如果有需要我替进行x-12-AMIMA季节调整的
收费,每个序列1000币.
我大概看了两天才掌握的.
如果是其他形式的调整,比如HP滤波,BP滤波或者指数平滑,那就便宜了,100币以内都可以搞定.

如果只有五年的月度数据，使用内置的5个ARIMA模型，最后三年的MAPE都在30%左右，你有社么建议？比如这些数据：
2008M01        17.419        12.979        13.219        13.653        19.142        29.027
2008M07        24.023        20.094        17.76        17.513        17.333        18.416
2009M01        13.335        16.893        21.677        21.943        22.555        18.413
2009M07        17.435        16.977        18.167        19.206        15.903        13.284
2010M01        14.326        24.361        29.829        32.193        26.535        23.543
2010M07        24.558        26.868        29.887        28.652        24.100        19.548
2011M01        17.787        27.307        42.832        56.267        43.019        34.447
2011M07        32.181        34.719        45.990        60.49        58.647        41.532
2012M01        36.761        40.983        54.490        61.383        55.242        41.647
2012M07        40.994        45.723        57.797        55.868        43.323        25.997

Model   1: (0 1 1)(0 1 1)

Average absolute percentage error in within-sample forecasts:
  Last year:  25.73      Last-1 year:  27.27     Last-2 year:  56.86
  Last three years:   36.62

  Chi Square Probability:    17.04%

  Nonseasonal MA parameter estimates:    -0.347
  Seasonal MA parameter estimates:        0.189


MODEL   1 REJECTED:
   Average forecast error >  15.00%


Model   2: (0 1 2)(0 1 1)

Average absolute percentage error in within-sample forecasts:
  Last year:  23.07      Last-1 year:  27.88     Last-2 year:  58.36
  Last three years:   36.44

  Chi Square Probability:    13.74%

  Nonseasonal MA parameter estimates:    -0.384  -0.047
  Seasonal MA parameter estimates:        0.196


MODEL   2 REJECTED:
   Average forecast error >  15.00%


Model   3: (2 1 0)(0 1 1)

Average absolute percentage error in within-sample forecasts:
  Last year:  30.16      Last-1 year:  24.89     Last-2 year:  40.65
  Last three years:   31.90

  Chi Square Probability:    51.70%

  Nonseasonal AR parameter estimates:     0.382  -0.342
  Seasonal MA parameter estimates:        0.251


MODEL   3 REJECTED:
   Average forecast error >  15.00%


Model   4: (0 2 2)(0 1 1)

Average absolute percentage error in within-sample forecasts:
  Last year:  70.68      Last-1 year:  52.65     Last-2 year: 132.03
  Last three years:   85.12

  Chi Square Probability:    22.74%

  Nonseasonal MA parameter estimates:     0.643   0.355
  Seasonal MA parameter estimates:        0.169


MODEL   4 REJECTED:
   Average forecast error >  15.00%
  Evidence of nonseasonal overdifferencing.


Model   5: (2 1 2)(0 1 1)

Average absolute percentage error in within-sample forecasts:
  Last year:  20.45      Last-1 year:  25.47     Last-2 year:  28.57
  Last three years:   24.83

  Chi Square Probability:    21.01%

  Nonseasonal AR parameter estimates:     1.006  -1.000
  Nonseasonal MA parameter estimates:     0.794  -0.708
  Seasonal MA parameter estimates:        0.985


MODEL   5 REJECTED:
   Average forecast error >  15.00%
  WARNING: Evidence of seasonal overdifferencing (see message below).


               None of the models were chosen.

1.不好意思，穷人，目前只有300来个论坛币，可以给你一部分。
2.这个操作过程我也基本了解，试了多次没有合适的模型，想听听您的建议。
谢谢。

你的ARIMA模型的参数设置太乱了,
目前最好不要自己随便设定
eviews默认的是(0 1 1) (0 1 1)
而且这种效果很多