適用於：
Databricks SQL Databricks Runtime 10.2 和以上版本
在expr上使用numBins個箱來計算直方圖。
語法
histogram_numeric ( [ALL | DISTINCT ] expr, numBins ) [ FILTER ( WHERE cond ) ]
引數
-
expr：函式取用的數值、TIMESTAMP、DATE或INTERVAL運算式，並計算其上的直方圖。 -
numBinsINTEGER：必須大於 1 的常值，指定直方圖計算的 bin 數目。 -
cond：選擇性BOOLEAN表達式，可篩選匯總的數據列。
退貨
傳回值是 ARRAYSTRUCTS 具有字段 x 和 y，用來表示直方圖分箱的中心。x 的型別與 expr 的型別相同，而 y 的型別是 DOUBLE。
增加numBins的值會改進直方圖的近似，使其更精細。 不過，它可能會在極端值周圍引入瑕疵。
一般而言，20-40 個箱對直方圖有效，不過扭曲或較小的數據集可能需要更多的箱。請注意，此功能會建立具有非統一箱寬的直方圖。
在直方圖的均方誤差方面，它不提供任何保證，但實際上與其他運算套件所產生的直方圖相當。
指定 DISTINCT 讓函式只能在一組唯一 expr 值上運作。
範例
> SELECT histogram_numeric(col, 5)
FROM VALUES (0), (1), (2), (10) AS tab(col);
[{"x":0.0,"y":1.0},{"x":1.0,"y":1.0},{"x":2.0,"y":1.0},{"x":10.0,"y":1.0}]
> SELECT histogram_numeric(col, 5)
FROM VALUES (0L), (1L), (2L), (10L) AS tab(col);
[{"x":0,"y":1.0},{"x":1,"y":1.0},{"x":2,"y":1.0},{"x":10,"y":1.0}]
> SELECT histogram_numeric(col, 5)
FROM VALUES (0F), (1F), (2F), (10F) AS tab(col);
[{"x":0.0,"y":1.0},{"x":1.0,"y":1.0},{"x":2.0,"y":1.0},{"x":10.0,"y":1.0}]
> SELECT histogram_numeric(col, 5)
FROM VALUES (0D), (1D), (2D), (10D) AS tab(col);
[{"x":0.0,"y":1.0},{"x":1.0,"y":1.0},{"x":2.0,"y":1.0},{"x":10.0,"y":1.0}]
> SELECT histogram_numeric(col, 5)
FROM VALUES (INTERVAL 0 YEAR), (INTERVAL 1 YEAR), (INTERVAL 2 YEAR),
(INTERVAL 3 YEAR) AS tab(col);
[{"x":0-0,"y":1.0},{"x":1-0,"y":1.0},{"x":2-0,"y":1.0},{"x":3-0,"y":1.0}]
> SELECT histogram_numeric(col, 5)
FROM VALUES (INTERVAL 0 DAY), (INTERVAL 1 DAY), (INTERVAL 2 DAY),
(INTERVAL 3 DAY) AS tab(col);
[{"x":0 00:00:00.000000000,"y":1.0},{"x":1 00:00:00.000000000,"y":1.0},{"x":2 00:00:00.000000000,"y":1.0},{"x":3 00:00:00.000000000,"y":1.0}]
> SELECT histogram_numeric(col, 5)
FROM VALUES (TIMESTAMP '2020-01-01'), (TIMESTAMP'2020-02-01'),
(TIMESTAMP'2020-03-01'), (TIMESTAMP'2020-10-01') AS tab(col)
[{"x":2020-01-01 00:00:00,"y":1.0},{"x":2020-02-01 00:00:00,"y":1.0},{"x":2020-03-01 00:00:00,"y":1.0},{"x":2020-10-01 00:00:00,"y":1.0}]