Declustering

Declustering your data involves applying weights to data points that are inversely proportional to the number of data points in a cell of a user-defined grid. Each cell is assigned a weight that is inversely proportional to the number of data points in that cell. This means that points from highly concentrated cells (clustered data) are weighted less than points from sparsely populated cells. The weights are then normalised to 1.

The decluster component in Supervisor creates a new column that contains the decluster weights, which can be analysed like any other column. This weight column can also be reused wherever you can select a weight field.

For a decluster component under an assay or a domain under an assay, the decluster weight field is automatically applied to the assay or domain and overwrites any weight field that may have been selected at the time of import. Any weight fields affected are disabled until the decluster component is disabled or removed.

Note: The weight column is not automatically used for child branches for decluster components under a Data Loader Component. You need to manually select this weight field for one or more of the assays in the component tree.