Best Practices for Filters

Filter on the Where tab first because the WHERE clause reduces the rows to consider for further subsetting.

When you specify a filter on a column that is an index or primary key, avoid using a function whenever it is possible. Using a function, such as CAST(order_id as DOUBLE), risks performing a full-table scan rather than using the index.

In some cases, you can improve performance by filtering on the table that has the smallest number of rows first.

If you must remove a table that is used on the Where or Having tabs, then remove the reference to the table from the filters before you remove the table. Otherwise, if you remove a table that is referenced in a filter, then all of the filtering conditions are cleared.

When adding a subquery using the Libraries tab, an additional step is needed if the column names or table names are written in a language that uses a double-byte character set. After you finish adding the columns, you must manually enter an n-literal to the string on the Code tab for each column name or table name that uses a double-byte character set. For example:

'table-name'n.'column-name'n

For more information about editing code manually, see Customizing Code.