Nearness settings

When finding similar records, you can edit the attributes and the criteria used to calculate 'nearness'. Nearness settings are saved with a selection project. When Find Similar is using settings that are not the default for the dataset, a warning is displayed.

Note: This feature is not available with Granta EduPack Introductory.

Click Find Similar, and select Nearness Settings.
By default, only the attributes that are populated for the current reference record will be shown. To show all available attributes, select Show all attributes. A warning icon will be shown next to the attributes that have no data for the reference record. To ensure accurate results, you should not include any attributes in the nearness calculation which are unpopulated.

Select the attributes to include in the nearness calculation, and choose how to use those attribute values in your calculation:

Option	Description
100% when	Sets how data values higher or lower than the reference record are treated in the nearness calculation. There are three options to choose from: Identical - Only having the same data value as a reference record is considered as no difference. Differences are calculated for values higher or lower than the reference record. Same or higher - Having the same data value or a higher data value than the reference record is considered as no difference. Differences are calculated for values lower than the reference record. Same or lower - Having the same data value or a lower data value than the reference record is considered as no difference. Differences are calculated for values higher than the reference record.
Weighting factor	The relative weighting that is given to an attribute in the nearness calculation. It must be between 0.01 and 100 (default is 1).

Option

Description

100% when

Sets how data values higher or lower than the reference record are treated in the nearness calculation. There are three options to choose from:

Identical - Only having the same data value as a reference record is considered as no difference. Differences are calculated for values higher or lower than the reference record.
Same or higher - Having the same data value or a higher data value than the reference record is considered as no difference. Differences are calculated for values lower than the reference record.
Same or lower - Having the same data value or a lower data value than the reference record is considered as no difference. Differences are calculated for values higher than the reference record.

Weighting factor

The relative weighting that is given to an attribute in the nearness calculation. It must be between 0.01 and 100 (default is 1).