Aggregate

QUANTILE_CONT

DuckDB

Menghitung kuantil kontinyu dengan interpolasi linear. Alias dari QUANTILE yang eksplisit menunjukkan metode continuous interpolation.

Tipe hasil: DOUBLE (atau LIST of DOUBLE jika input quantile adalah array)Diperbarui: 6 Jan 2026

Syntax

SQL

QUANTILE_CONT(expression, quantile) | QUANTILE_CONT(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

Continuous Percentile Calculation

SQL

1 SELECT
2   department,
3   QUANTILE_CONT(salary, 0.5) as median_salary,
4   QUANTILE_CONT(salary, 0.9) as p90_salary
5 FROM employees
6 GROUP BY department;

Menghitung median dan p90 salary dengan interpolasi kontinyu.

Hasil

department	median_salary	p90_salary
IT	17500000.00	32500000.00
Sales	12750000.00	28000000.00
HR	10250000.00	18500000.00

Interpolated Quartiles

SQL

1 SELECT
2   product_type,
3   QUANTILE_CONT(rating, [0.25, 0.5, 0.75]) as rating_quartiles
4 FROM product_reviews
5 GROUP BY product_type;

Menghitung IQR (Interquartile Range) untuk rating produk.

Hasil

product_type	rating_quartiles
Electronics	[3.75, 4.20, 4.65]
Books	[3.50, 4.00, 4.50]

Performance Monitoring dengan Continuous Percentiles

SQL

1 SELECT
2   service_name,
3   ROUND(QUANTILE_CONT(latency_ms, 0.50), 2) as p50,
4   ROUND(QUANTILE_CONT(latency_ms, 0.95), 2) as p95,
5   ROUND(QUANTILE_CONT(latency_ms, 0.99), 2) as p99
6 FROM service_metrics
7 WHERE timestamp >= NOW() - INTERVAL '1 hour'
8 GROUP BY service_name
9 HAVING QUANTILE_CONT(latency_ms, 0.99) > 100;

Monitoring service latency dengan alert jika p99 > 100ms.

Hasil

service_name	p50	p95	p99
auth-service	15.25	78.50	125.75
api-gateway	22.10	95.25	180.50

Kembali ke DuckDB Functions

1	SELECT
2	department,
3	QUANTILE_CONT(salary, 0.5) as median_salary,
4	QUANTILE_CONT(salary, 0.9) as p90_salary
5	FROM employees
6	GROUP BY department;

1	SELECT
2	product_type,
3	QUANTILE_CONT(rating, [0.25, 0.5, 0.75]) as rating_quartiles
4	FROM product_reviews
5	GROUP BY product_type;

1	SELECT
2	service_name,
3	ROUND(QUANTILE_CONT(latency_ms, 0.50), 2) as p50,
4	ROUND(QUANTILE_CONT(latency_ms, 0.95), 2) as p95,
5	ROUND(QUANTILE_CONT(latency_ms, 0.99), 2) as p99
6	FROM service_metrics
7	WHERE timestamp >= NOW() - INTERVAL '1 hour'
8	GROUP BY service_name
9	HAVING QUANTILE_CONT(latency_ms, 0.99) > 100;