![]() |
Generally, you will want to create a predicate column using one of the comparison macro functions (for example, ==, >, <, ISEVEN, ISODD, etc.). You can then extract the rows of interest from a specified data range using the EXTRACT macro function. This can be useful to "weed out" bad data points (for example, when a particular value exceeds the maximum or minimum value for a data variable). It also can be used to consolidate all examples of a particular class (for example, if the column V3 contains ones and zeros for one of the output classes, use V4=EXTRACT(V3, V1:V2) to extract the inputs V1 and V2). Since EXTRACT condenses all the extracted rows as a block of data (that is, it fills the cell range VX[1:n}:VY), where n is the number of extracted rows, it is a useful function for copying a range of cells from their current row locations to rows 1- n of the spreadsheet.
|
Creates two new columns named TEMP and VX with values in the first 51 cells. The values in the TEMP column are the cells 50-100 of column V1, and the values in the VX column are the cells 50-100 of column V2.
|
Creates two new columns named TEMP and VX. For each row where the value in column V3 is one, the corresponding row across columns V1 and V2 are extracted into columns TEMP and VX, respectively.
|
Creates a new column named TEMP containing all the values in column V1 that were greater than the corresponding values in column V2.
|
Creates two new columns named TEMP and VX. For rows 10-20 where the value in column V3 is one, the corresponding row across columns V1 and V2 is extracted into columns TEMP and VX, respectively.
|
Copyright IBM Corporation 2012. All Rights Reserved.
|