AggHAR Command Line Aggregation program

AggHAR is a specialized program which offers some aggregation features that ViewHAR lacks. For example, unlike ViewHAR, AggHAR is able to perform weighted aggregation of parameters (see aggregation feature comparison). It uses double-precision arithmetic for greater accuracy.

AggHAR is a command line program which you must run from a commmand prompt (aka DOS box). You might type:

AggHAR OLD.HAR NEW.HAR SUP.HAR -PM

above:

OLD.HAR is the input file to be aggregated.

NEW.HAR is the new output file (must NOT exist yet).

SUP.HAR is another existing HAR file containing mappings.

If the final, options, item is present, it must begin with a hyphen and contain some or all the following letters:

U discard headers Unaffected by aggregation - see

P use Previous set names for new headers - see

M conserve Memory - see

You must put aggregation information in the supplementary file (SUP.HAR above). The necessary information is:

(a) String headers specifying the new, aggregated sets. Use ViewHAR's Create Set command to make these.

(b) Integer or string headers specifying the mapping from the old sets to the new sets. Use ViewHAR's Create Mapping command to make these.

(c) If you want weighted aggregation you must specify a list of headers to be aggregated together with a list of the corresponding weight matrices. A special format is used.

Try out the AggHAR example.

How AggHAR works

You might find the following information helpful in interpreting AggHAR output. The program goes through the following stages:

•checks command line syntax and existence of input files.

•extracts from the main input file details of old sets which are used to dimension real arrays and so are candidates for remapping.

•extracts set information from real and string headers on the supplementary input file. These would normally include the aggregated sets.

•deduces which sets are subsets of other sets.

•reads in any mappings from the supplementary input file. A mapping will only be recognized if the two sets to which the mapping refers have already been recognized.

•lists each set found in the main input file, and reports what mapping, if any, can be used to aggregate it. At this stage AggHAR may infer mappings.

•reads from the supplementary input file the list of headers for weighted aggregation.

•does the aggregation, reporting on each header in turn. 1C (string) headers on the old file that seem to describe sets which are to be remapped are replaced with new 1C headers describing the corresponding target set.

•reports final success or failure, and time taken.

Problems

AggHAR relies on you to store new sets and mappings on the SUPP file in the right way -- so that they can be recognized as sets or mappings. The long name or description field must be specified correctly: this should happen if you use ViewHAR's Create Set and Create Mapping commands to make the sets and mappings. It will also happen if the SUPP file is generated by a TABLO program. See Long Names of Mapping Vectors for more information.

One way to check a set or mapping is: from ViewHAR contents page, right-click on the header and choose Properties -- a message may tell you if and why that header was recognized as a set or mapping.

If your database contains shares you may encounter difficulties in using AggHAR to aggregate them, especially if you aggregate 2 dimensions at once. See Problems aggregating shares.

URL of this topic: www.copsmodels.com/webhelp/viewhar/hc_agghar.htm

Link to full GEMPACK Manual

Link to GEMPACK homepage