Aggregation

The following are scalar functions that you can use in the stats and aggregate with trigger streaming functions to perform calculations over data in a given time-window.

average(value)

Calculates the average (mean) of values in a time window.

Function Input

value: T

Function Output

double

SPL2 example

The following example returns the average (mean) "size" for each distinct "host".

When working in the SPL View, you can write the function by using the following syntax.

...| stats average(size) BY host, span(timestamp, 50s, 10s) |...; 

Alternatively, you can use named arguments.

...| stats average(value: size) BY host, span(timestamp, 50s, 10s) |...; 

count(value)

Returns the number of non-null values in a time window.

Function Input

value: any

Function Output

long

SPL2 example

Returns the count of the "status_code" field.

When working in the SPL View, you can write the function by using the following syntax.

...| stats count(status_code) by status_code, span(window_start, 5000ms, 1000ms, 1000ms) |...;

Alternatively, you can use named arguments.

...| stats count(value: status_code) by status_code, span(window_start, 5000ms, 1000ms, 1000ms) |...;

estdc(value)

Estimated Distinct Count (estdc) is a stats function that calculates an approximated distinct count value for any field. This function works with ~1.5% error bound.

Function Input

value: string

Function Output

long

SPL2 example

Returns an estimated count of the distinct values in the input field.

When working in the SPL View, you can write the function by using the following syntax.

... | stats estdc(input) by span(timestamp, 10ms);

Alternatively, you can use named arguments.

... | stats estdc(value: input) by span(timestamp, 10ms);

max(value)

Returns the maximum value in a time window.

Function Input

value: number

Function Output

number

SPL2 example

Returns the maximum value of the "time_taken" field.

When working in the SPL View, you can write the function by using the following syntax.

...| stats max(time_taken) by time_taken, span(timestamp, 50s, 10s) |...;

Alternatively, you can use named arguments.

...| stats max(value: time_taken) by time_taken, span(timestamp, 50s, 10s) |...;

mean(value)

Calculates the average (mean) of values in a time window.

Function Input

value: number

Function Output

double

SPL2 example

Returns the average value of the "time_taken" field.

When working in the SPL View, you can write the function by using the following syntax.

...| stats mean(time_taken) by time_taken, span(timestamp, 50s, 10s) |...;

Alternatively, you can use named arguments.

...| stats mean(value: time_taken) by time_taken, span(timestamp, 50s, 10s) |...;

min(value)

Returns the minimum value in a time window.

Function Input

value: number

Function Output

number

SPL2 example

Returns the minimum value of the "time_taken" field.

When working in the SPL View, you can write the function by using the following syntax.

...| stats min(time_taken) by time_taken, span(timestamp, 50s, 10s) |...;

Alternatively, you can use named arguments.

...| stats min(value: time_taken) by time_taken, span(timestamp, 50s, 10s) |...;

perc(value, q1, q2)

Percentiles (perc) is a stats function that computes the approximate q-th percentile value of a numeric field input field with ~1.5% error bound. The perc(input, 0.25, 0.5, 0.75) takes multiple quantile queries as inputs, and then outputs fields as a list with appropriate percentile values.

Function Input

value: number

q1: double

q2: double

Function Output

collection: double

SPL2 examples

When working in the SPL View, you can write the function by using the following syntax.

...| stats perc(input, 0.75) by span(timestamp, 10ms);

...| stats perc(input, 0.5, 0.99) by span(timestamp, 10ms);

sum(value)

Returns the sum of values in a time window.

Function Input

value: number

Function Output

number

SPL2 example

Returns the sum of the "time_taken" field.

When working in the SPL View, you can write the function by using the following syntax.

...| stats sum(time_taken) by time_taken, span(timestamp, 50s, 10s) |...;

Alternatively, you can use named arguments.

...| stats sum(value: time_taken) by time_taken, span(timestamp, 50s, 10s) |...;