isfinite

isfinite is essential for data quality checks. Division by zero, logarithm of a non-positive number, or other edge cases can produce infinity or NaN in your computed columns. Use isfinite to identify or filter these invalid values before aggregating or visualizing data.

Usage#

Syntax#

isfinite(x)

Parameters#

Name	Type	Required	Description
`x`	real	Yes	The numeric value to check.

Returns#

true if x is a finite real number (not infinity and not NaN). false otherwise.

Example#

Use isfinite to check whether a computed value is finite before using it in aggregations.

Query

['sample-http-logs']
| extend log_duration = log(req_duration_ms)
| extend is_valid = isfinite(log_duration)
| project _time, id, req_duration_ms, log_duration, is_valid

Run in Playground

Output

_time	id	req_duration_ms	log_duration	is_valid
2024-11-14 10:00:00	user-1	120.0	4.7875	true
2024-11-14 10:01:00	user-2	0.0	-inf	false
2024-11-14 10:02:00	user-3	80.0	4.3820	true

isinf: Returns true only for infinite values. Use it when you want to distinguish infinity from NaN.
isnan: Returns true only for NaN values. Use it when you want to detect only Not-a-Number results.
isint: Returns true for integer values. Use it to check whether a numeric value is an integer rather than a floating-point result.
log: Returns the natural logarithm. Combine with isfinite to validate log computations on potentially non-positive inputs.
gamma: Returns the gamma function. Use isfinite to filter out overflow results when computing gamma on large values.

Other query languages#

Splunk SPL users

Splunk SPL doesn't have a direct isfinite() function. You typically check for null or use isnum() to verify that a value is a valid number, but these checks don't cover infinity.

Splunk example

| eval is_valid = if(isnum(value) AND value != 'Infinity', 1, 0)

APL equivalent

['sample-http-logs']
| extend is_valid = isfinite(value)

ANSI SQL users

Standard SQL does not define an ISFINITE() function. You typically check for IS NOT NULL and whether the value falls within a valid range, but you cannot directly detect infinity in most SQL dialects.

SQL example

SELECT CASE WHEN value IS NOT NULL AND value BETWEEN -1e308 AND 1e308 THEN 1 ELSE 0 END AS is_valid FROM logs