Aggregating or Remapping Data

Overview

ViewHAR is able to aggregate matrices of real numbers. For example, a vector of 50 industry wage bills could be viewed as a vector of wages in 10 broad industrial groups. Each of the 50 original industries must correspond to just one broad group.

The mappings are read from a HAR file, where they are stored in a special way. You choose which mappings are active. Suppose you activated a mapping from the set IND (50 industries) to the set BROADIND (10 broad sectors) You could create a second view of the current data file where IND-sized dimensions of all real matrices were replaced with BROADIND dimensions.

You can toggle between the two views. You can change which mappings are active: several mappings may be active simultaneously. You can save the aggregated data as a new file.

To learn more, follow the Aggregation Example tutorial.

It is anticipated that in most cases the mappings that are used will have been written to a HAR file by a TABLO-generated program. However, you can also create new mappings within ViewHAR. If you want to create your own mappings by either method, you will need to have intermediate level of GEMPACK knowledge and to read all relevant topics in this help file very carefully. Use the file that comes with the Aggregation Example as a pattern.

A mapping specifies a source set and a target set; and, for each member of the source set, the corresponding element in the target set. Normally:

(a) the source set has more elements than the target set.

(b) each member of the target set is mapped to by at least one member of the source set.

However ViewHAR does not insist on these conditions. You may exploit this to reorder or expand sets. See Strange mappings.

A more specialized program, AggHAR, offers some additional aggregation features. See AggHAR.