Data adjustment

Plaid has a limited facility to make adjustments on the data. Adjustments can be made to the rows or to the columns or to both. A location adjustment visits each entity (row or column or both) in turn, computes a value for that entity, and subtracts it from every data element in the entity. A scale adjustment divides each data element by a value. When both location and scale adjustments are requested, the location adjustment is done first then the scale adjustment.

The command adjust carries this out. Here are some examples of this command:

adjust row scale moment
adjust col location robust
adjust both location moment
adjust col both robust
adjust both both moment 10

The first argument to adjust describes whether rows, columns or both are to be adjusted. The second column describes whether the adjustment should be for location, or scale, or both. The third argument can be moment for which location and scale are mean and standard deviation, or robust for median and median absolute deviation, or extreme for minimum and maximum minus minimum.

A fourth argument, if present, describes the number of times to apply the adjustment if this number is to be larger than $1$. For example, to make every row have the same scale and every column have the same scale may require iteration.

If two arguments are given, then one moment adjustment is made. If one argument is given, then one scale moment adjustment is made. If no arguments are given, then one column scale moment adjustment is made. The algorithm will refuse to divide any row or column by $0$, and will instead print a warning. Any number of adjust commands can be made. They are applied in sequence.