Filtered aggregation functions

Comparative filters compare two segments of some whole against each in the Search bar. For example, a company that has locations across the United States, may want to compare total revenue in the West to the East segment. In a comparative filter, one of the segments you are comparing is filtered.

An example of a comparative filter is comparing west revenue with total revenue. In these cases, one measure is a filtered measure, for example, revenue region = west is a filtered measure.

Derivative filters add a column to your results which is derived from other columns in the same results. For example, you search for revenue and cost and want to calculate profit in your result.

Some examples of comparative and derivative filters in the real world are:

If you plan to create these types of filters, you need to understand how to create filter functions.

Filter functions take two arguments, the column ( measure or attribute) to aggregate and the filter condition:

ThoughtSpot functional library will include the following functions:

Consideer the following examples of these functions:

A condition can have multiple filters, like the two conditions for region in the example sales region = west OR region = east.

You can also just type a value as a filter; here, east is the filter in the expression sales east.

If there are no rows matching the filter criteria, the condition returns a 0 (zero). When you use a filter rather than a filter function, ThoughtSpot does not include any zero values in your chart or table. Using a filter function like sum_if or count_if is the best way to include zero values in your chart or table.

A 0 can result in situations where there are logic errors in the formula, so be sure to verify your work.

After you have aggregated with a filter, you can do further comparisons with the vs and all keywords.

You can use the vs and all keywords to expand the usefulness of your comparison filters. It compares a measure across different sets of filters and or groupings. The basic format of a comparison search is:

<common search tokens> (A vs B) <common search tokens>

For example:

revenue region last 10 years vs all

Try this syntax on using the superstore example data. The first vs example compares two segments with a single search token:

The system automatically applies the sales token to both sides and groups each segment. You can use the all keyword to break out the segments and avoid grouping.

You can also provide multiple vs instances:

Of course, you can compare across different columns as well:

Other supported formats you can try: