Aggregate

QUANTILE

DuckDB

Menghitung nilai pada posisi kuantil tertentu dari distribusi data. Mendukung single quantile atau array of quantiles sekaligus.

Tipe hasil: Sama dengan tipe input (atau LIST jika input quantile adalah array)Diperbarui: 6 Jan 2026

Syntax

SQL

QUANTILE(expression, quantile) | QUANTILE(expression, [q1, q2, ...])

Parameter

expressionnumericwajib

Kolom atau ekspresi numerik yang akan dihitung kuantilnya

quantileDOUBLE atau LISTwajib

Posisi kuantil (0.0 sampai 1.0) atau array of quantiles

Contoh Penggunaan

Menghitung Quartiles

SQL

1 SELECT
2   category,
3   QUANTILE(price, 0.25) as Q1,
4   QUANTILE(price, 0.50) as median,
5   QUANTILE(price, 0.75) as Q3
6 FROM products
7 GROUP BY category;

Menghitung quartile 1, median, dan quartile 3 harga per kategori.

Hasil

category	Q1	median	Q3
Electronics	500000	2500000	8000000
Clothing	150000	350000	750000

Multiple Quantiles dengan Array

SQL

1 SELECT
2   region,
3   QUANTILE(salary, [0.1, 0.5, 0.9]) as salary_distribution
4 FROM employees
5 GROUP BY region;

Menghitung persentil 10, 50, dan 90 sekaligus dengan array.

Hasil

region	salary_distribution
Jakarta	[8000000, 15000000, 35000000]
Surabaya	[6000000, 12000000, 25000000]

Percentile untuk Response Time

SQL

1 SELECT
2   endpoint,
3   QUANTILE(response_ms, 0.50) as p50,
4   QUANTILE(response_ms, 0.90) as p90,
5   QUANTILE(response_ms, 0.99) as p99,
6   MAX(response_ms) as max_response
7 FROM api_logs
8 WHERE log_date = CURRENT_DATE
9 GROUP BY endpoint
10 ORDER BY p99 DESC;

Analisis latency dengan p50, p90, dan p99 untuk monitoring SLA.

Hasil

endpoint	p50	p90	p99	max_response
/api/search	45	120	350	1250
/api/users	12	35	85	320
/api/health	2	5	10	45

Kembali ke DuckDB Functions

1	SELECT
2	category,
3	QUANTILE(price, 0.25) as Q1,
4	QUANTILE(price, 0.50) as median,
5	QUANTILE(price, 0.75) as Q3
6	FROM products
7	GROUP BY category;

1	SELECT
2	region,
3	QUANTILE(salary, [0.1, 0.5, 0.9]) as salary_distribution
4	FROM employees
5	GROUP BY region;

1	SELECT
2	endpoint,
3	QUANTILE(response_ms, 0.50) as p50,
4	QUANTILE(response_ms, 0.90) as p90,
5	QUANTILE(response_ms, 0.99) as p99,
6	MAX(response_ms) as max_response
7	FROM api_logs
8	WHERE log_date = CURRENT_DATE
9	GROUP BY endpoint
10	ORDER BY p99 DESC;