The filter factor is a ratio (a number between 0 and 1) that estimates I/O costs. It is an estimate of the proportion of table rows for which a predicate is true. The formulas used by the optimizer to calculate the filter factor are proprietary IBM information.
These formulas assume uniform data distribution, so they should be used only when determining the filter factor for static SQL queries or queries on tables having no distribution statistics stored in the DB2 Catalog. The filter factor for dynamic SQL queries is calculated using the distribution statistics, in SYSCOLDIST , if available.
Filter Factor is the percentage of rows that qualify for a coded predicate. It determines if an index will be used.