Applying a filter

You usually can drag a discrete value directly to a filter area. For example, in an analysis of recent orders, you could drag the Customer Gender discrete value “female” to the analysis filter to see only orders placed by female customers.

A universal filter is applied before any change in resolution occurs. A target filter is applied after a change in resolution occurs. For example, to view only records for female customers, add as a universal filter: Gender equals female. If you add Gender equals female as a target filter and change the resolution from Customer to Household, only records that include households with females appear. Some of those households may also include males.

Target and baseline filters are used together to create comparative analyses. Be sure to use segments that can be compared. For example, compare one year with another or one population group with another. When calculating a comparative analysis, you can choose to display a measure as:

The default for measures is to produce the count of records in both filtered segments. This is not directly useful for most comparisons but can be used as a total when creating calculated fields.

Shows the degree to which compared groups differ using an indicator. An index value greater than 0 means that the baseline is as many times greater than the value shown by the index with respect to the target. An index value less than 0 means the reverse is true. The formula for Index is:

Displays size differences between the baseline and the target as measured in units. A negative result means that the baseline has as many more values than the displayed number. A positive result indicates the opposite. The formula for Difference is:

If you are using filters, note that specific situations require certain filter types. You must use a target filter for pivoted analyses when there is a change in resolution between the axes and the measures in the direction N-to-1. You must use a universal filter with a non‑pivoted table when there is a change in resolution between the axes and the measures in the direction N‑to‑1. When no such size disparity exists between axes and measures, the type of filter used for each analysis does not matter.

Universal filter. Apply the filter before carrying out the change in resolution. For example, the field low salary [axis 1: customer table, salary field] is used with the filter. Then, change the resolution to a different table, such as Households. The filter conditions are met by the same household and person. The resolution unit of the filter is the unit indicated by the axes of the crosstab, in this case Customers.

Target filter. Apply the filter after carrying out the change in resolution. In other words, select the segments from the table to which the selected axes belong, and carry out the change in resolution for a table chosen as the resolution level for the results. Then, apply the filter. For example, low salary [axis 1: customer table, salary field] and the resolution is changed to Household. In this example, you see all households with at least one customer whose salary is low. The filter is applied. For example, [customer tables, gender field = female] gives a result qualitatively higher than the result from the universal filter. All households meeting the condition of low salary and female appear. The condition is not necessarily met by the same person. The filter's resolution unit is Households, the resolution table for the crosstab.

Baseline filter. Selecting a target filter activates a baseline filter. Use a baseline filter for the purpose of comparison. For example, compare two periods of time using the following two filters: 2008 target and 2007 baseline.

To configure the table, you must first select the axis or axes by dragging to the appropriate space, then dragging the measures. By default, when you drag the axes, the value count for the table to which they belong appears.

Parameters used in filters should be included if a table is calculated or in a situation where you introduce a new data table in order to calculate the final output. The filter is determined by prompting you for the value when the analysis is calculated. You can use either a pre-set filter or a prompted filter, but not both.