top-hitters operator

  • Article

Returns an approximation for the most popular distinct values, or the values with the largest sum, in the input.

Note

top-hitters uses an approximation algorithm optimized for performance when the input data is large. The approximation is based on the Count-Min-Sketch algorithm.

Syntax

T | top-hitters NumberOfValues of ValueExpression [ by SummingExpression ]

Learn more about syntax conventions.

Parameters

Name Type Required Description
T string ✔️ The input tabular expression.
NumberOfValues int, long, or real ✔️ The number of distinct values of ValueExpression.
ValueExpression string ✔️ An expression over the input table T whose distinct values are returned.
SummingExpression string If specified, a numeric expression over the input table T whose sum per distinct value of ValueExpression establishes which values to emit. If not specified, the count of each distinct value of ValueExpression is used instead.

Remarks

The first syntax (no SummingExpression) is conceptually equivalent to:

T | summarize C``=``count() by ValueExpression | top NumberOfValues by C desc

The second syntax (with SummingExpression) is conceptually equivalent to:

T | summarize S``=``sum(*SummingExpression*) by ValueExpression | top NumberOfValues by S desc

Examples

Get most frequent items

The next example shows how to find top-5 types of storms.

Run the query

StormEvents
| top-hitters 5 of EventType

Output

EventType approximate_count_EventType
Thunderstorm Wind 13015
Hail 12711
Flash Flood 3688
Drought 3616
Winter Weather 3349

Get top hitters based on column value

The next example shows how to find the States with the most "Thunderstorm Wind" events.

Run the query

StormEvents
| where EventType == "Thunderstorm Wind"
| top-hitters 10 of State

Output

State approximate_sum_State
TEXAS 830
GEORGIA 609
MICHIGAN 602
IOWA 585
PENNSYLVANIA 549
ILLINOIS 533
NEW YORK 502
VIRGINIA 482
KANSAS 476
OHIO 455