quantileBFloat16Weighted

Like quantileBFloat16 but takes into account the weight of each sequence member.

Computes an approximate quantile of a sample consisting of bfloat16 numbers.

bfloat16 is a floating-point data type with 1 sign bit, 8 exponent bits and 7 fraction bits. The function converts input values to 32-bit floats and takes the most significant 16 bits. Then it calculates bfloat16 quantile value and converts the result to a 64-bit float by appending zero bits. The function is a fast quantile estimator with a relative error no more than 0.390625%.

Syntax

quantileBFloat16Weighted(level)(expr, weight)

Parameters

• level — Optional. Level of quantile. Possible values are in the range from 0 to 1. Default value: 0.5. Float*

Arguments

• expr — Column with numeric data. (U)Int* or Float*
• weight — Column with weights of sequence members. (U)Int* or Float*

Returned value

Approximate quantile of the specified level. Float64

Examples

Computing weighted quantile with bfloat16

CREATE TABLE example_table (a UInt32, b Float32, w UInt32) ENGINE = Memory;
INSERT INTO example_table VALUES (1, 1.001, 1), (2, 1.002, 2), (3, 1.003, 3), (4, 1.004, 4);

SELECT quantileBFloat16Weighted(0.75)(a, w), quantileBFloat16Weighted(0.75)(b, w) FROM example_table;

┌─quantileBFloat16Weighted(0.75)(a, w)─┬─quantileBFloat16Weighted(0.75)(b, w)─┐
│                                    3 │                                    1 │
└──────────────────────────────────────┴──────────────────────────────────────┘

Introduced in version 21.10.