SAQL Statements

A query is made up of statements. Each SAQL statement has an input stream, an operation, and an output stream.

arimax

Uses existing data to predict future data points. The arimax statement must follow a projection statement in your query. Perform any filtering pre-projection or after the arimax statement.
cogroup

Use cogroup to blend data from two or more data streams into a single data stream. The data streams must have at least one common field.
fill

Use fill() to fill in any gaps in date fields. You most often use fill() before using the timeseries statement. By specifying the date fields to check, fill() creates a row that contains the missing month, day, week, quarter, or year and includes a null value. To include values outside the bounds of your data’s date range, specify a start date and end date to override existing limits. The function returns the missing date rows with null values.
filter

Selects rows from a dataset based on a filter predicate.
foreach

Applies a set of expressions to every row in a dataset. This action is often referred to as projection.
group-by

Organizes the rows returned from a query into groups. Within each group, you can apply an aggregate function, such as count() or sum() to get the number of items or sum, respectively.
group-by rollup

rollup is a subclause of group-by that creates and displays aggregations of grouped data. The output of rollup is based on column order in your query.
join semi and anti

Use the join statement with the join_type to create semi-join or anti-join results.
limit

Limits the number of results that are returned. If you don’t set a limit, queries return a maximum of 10,000 rows.
load

Loads a dataset. All SAQL queries start with a load statement.
offset

Use offset to page through the results of your query.
order

Sorts in ascending or descending order on one or more fields.
sample

Returns a random sample from a large dataset, where each data point has an equal probability of being selected. This keyword uses the Bernoulli distribution.
timeseries

Uses existing data to predict future data points. The timeseries statement must follow a projection statement in your query. Perform any filtering pre-projection or after the timeseries statement.
union

Combines multiple result sets into one result set. The result sets must have the same field names and structure. You can use a different dataset to create each result set, or you can use the same dataset.