CategoryStatistics

Category

Transformation

Function

Calculate statistics on data associated with a categorical component

Syntax

statistics = CategoryStatistics(input, operation, category, data, lookup);

Inputs

Name	Type	Default	Description
input	field	(none)	field for which to compute statistics
operation	string	"count"	operation to perform ("count", "mean", "sd", "var", "min", "max")
category	string	"data"	component with categorical values
data	string	"data"	data component for statistics
lookup	integer, string, value list	"category lookup"	lookup component

Outputs

Name	Type	Description
statistics	field	field with data containing the statistics and positions for the category values

Functional Details

input	field containing the categorical and data components
operation	calculation to perform
category	component with categorical values. This component must be an integer type (int, ubyte, ...)
data	data component for statistics. This component must be scalar.
lookup	lookup component (optional)

CategoryStatistics calculates statistics on a scalar component associated with a categorical component. If the operation is "count", the data component is ignored and the number of counts in each category is calculated, corresponding to a histogram of the unique values in the categorized component.

For example, if input is a Field with component "state" containing the entries {1,0,1,2,3}, component "state lookup" containing the entries {"CA", "NY", "PA", "VA"}, and a component "sales" containing the entries {1.2,1.0,1.4,1.7,1.8}, then CategoryStatistics(input,"mean","state","sales") will produce an output field where the "positions" component will contain the indices {0,1,2,3} and the "data" component will contain the mean value for sales for each state, that is {1.0,1.3,1.7,1.8}.

The output of CategoryStatistics is a field with a "positions" component corresponding to the categorical indices, and a "data" component corresponding to the requested statistics. The "positions" component will consist of the integers 0 to N-1, where N can be determined in a number of ways:

• If no lookup component is specified, and if a "categoryname lookup" component is not found, (where "categoryname" is the string specified by category), then the output field will simply have positions from 0 to MAX_N, where MAX_N is the maximum integer found in the category component.
• If, on the other hand, a "categoryname lookup" component is found, or lookup is specified, then the number of category bins will be the number of items in lookup. lookup can also simply be an integer specifying the number of category bins.
• If a lookup table is provided, then for convenience, a "categoryname lookup" component will be placed in the output containing the values corresponding to the categorical indices.

Components

Creates an output field with a "positions" component representing the categorical indices, and a "data" component containing the requested statistics. Creates a "categoryname lookup" component if a lookup table is specified using the lookup parameter.

Example Visual Programs

Duplicates.net
Zipcodes.net

See Also

Categorize, Statistics, Lookup