Subsections:

The ESM procedure can be used to forecast time series data as well as transactional data. If the data is transactional, then the procedure must first accumulate the data into a time series before it can be forecast. The procedure uses the following sequential steps to produce forecasts, with the options that control the step listed to the right:

Table 14.2: ESM Processing Steps and Control Options

Step |
Operation |
Option |
Statement |
---|---|---|---|

1 |
accumulation |
ACCUMULATE= |
ID |

2 |
missing value interpretation |
SETMISSING= |
ID, FORECAST |

3 |
transformations |
TRANSFORM= |
FORECAST |

4 |
parameter estimation |
MODEL= |
FORECAST |

5 |
forecasting |
MODEL=, LEAD= |
FORECAST, PROC ESM |

6 |
inverse transformation |
TRANSFORM, MEDIAN |
FORECAST |

7 |
summation of forecasts |
LEAD=, STARTSUM= |
PROC ESM |

Each of the steps shown in Table 14.2 is described in the following sections.