Coclustering Rules
Coclustering Rules
A coclustering model is stored using a DataGrid structure. With a DataGridDeployment structure,
it can be deployed on a given dimension variable, for a distribution conditional on the values on
the rest coclustering variables. Other deployment rules can then be used to obtain the closest
part for the deployed variable, the distance to all the other parts on the deployed variable,
and to compute the aggregated frequencies on the parts of all the input variables.
DataGridDeployment
Structure(DataGridDeployment) DataGridDeployment(
Structure(DataGrid) dataGrid, Numerical deployedVariableIndex,
Vector inputValues1, Vector inputValues2, ..., [Vector inputFrequencies]
)
Builds a DataGridDeployment structure for a given deployment variable of a data grid. The input
values (Vector for Numerical values and VectorC for Categorical values) correspond to the
distribution of the input values. The frequency vector is optional.
PredictedPartIndex
Computes the index of the closest part of the deployed variable.
PredictedPartDistances
Computes the distance to all the parts of the deployed variable.