Last updated: Jan 17, 2024
Treating categorical data as continuous can have a serious impact on the quality of a
model, especially if it's the target field (for example, producing a regression model rather than a
binary model). To prevent this, you can convert integer ranges to categorical types such as
Ordinal
or Flag
.
- Double-click a Type node to open its properties. Expand the Type Operations section.
- Specify a value for Set continuous integer field to ordinal if range less than or equal to.
- Click Apply to convert the affected ranges.
- If desired, you can also specify a value for Set categorical fields to None if they exceed this many values to automatically ignore large sets with many members.
Results of the conversion
- Where a
Continuous
field with integer storage is changed toOrdinal
, the upper value is expanded to include all of the integer values through the upper value. For example, if you set the value to5
, the set of values is1, 2, 3, 4, 5
. - If the
Continuous
field changes toFlag
, the lower and upper values become the false and true values of the flag field.