arg_min

Output column naming follows Microsoft Kusto:

• The single alias (or the source column name if unaliased) is used only for the key column. loser = arg_min(score, name) → columns loser:long, name:string.
• Each value column is named after the underlying source column when the expression is a plain column reference or a simple unary call (todouble(v), bin(v, 1) → v); otherwise it is named min_<keyname>_arg<N> (1-based value-arg index).
• Tuple assignment names each output column explicitly: (min_score, loser) = arg_min(score, name) → min_score:long, loser:string.
• * expands to every non-key column with its own name and type.

Duplicate names across multiple aggregates are disambiguated by appending 1, 2, …: arg_max(score, name), arg_min(score, name) → score, name, score1, name1.

Output column types match the inferred types of each argument expression.

Syntax

arg_min(minimize, return_expr)

Parameters

Returns: any

Syntax

arg_min(minimize, return_expr)

Parameters

Returns: any

Examples

Example 1

datatable(clan:string, warrior:string, voyages:long)[
  "Lothbrok", "Ragnar", 42,
  "Lothbrok", "Bjorn", 31,
  "Fairhair", "Harald", 25,
  "Lothbrok", "Ivar", 35
]
| summarize arg_min(voyages, warrior) by clan