R, Python 분석과 프로그래밍의 친구 (by R Friend)

DROP TABLE IF EXISTS cust;

CREATE TABLE cust (id INT, age INT, gender TEXT);

INSERT INTO cust VALUES

(1,NULL,'M'),

(2,NULL,'M'),

(3,25,'M'),

(4,28,'M'),

(5,27,'M'),

(6,25,'M'),

(7,26,'M'),

(8,29,'M'),

(9,25,'M'),

(10,27,'M'),

(11,NULL,'F'),

(12,23,'F'),

(13,25,'F'),

(14,23,'F'),

(15,24,'F'),

(16,26,'F'),

(17,23,'F'),

(18,24,'F'),

(19,22,'F'),

(20,23,'F');

 (1) SQL로 연속형 데이터의 그룹별 요약통계량 구하기

SELECT

    gender AS group_by_value

    , 'age' AS target_column

    , COUNT(*) AS row_count

    , COUNT(DISTINCT age) AS distinct_values

    , AVG(age)

    , VARIANCE(age)

    , STDDEV(age)

    , MIN(age)

    , PERCENTILE_DISC(0.25) WITHIN GROUP (ORDER BY age) AS first_quartile

    , MEDIAN(age)

    , PERCENTILE_DISC(0.75) WITHIN GROUP (ORDER BY age) AS third_quartile

    , MAX(age)

    , MODE() WITHIN GROUP (ORDER BY age) -- over PostgreSQL 9.4

FROM cust

WHERE age IS NOT NULL

GROUP BY gender

ORDER BY gender;

SELECT 

    gender

    , COUNT(*) AS missing_count

FROM cust

WHERE age IS NULL

GROUP BY gender

ORDER BY gender;

WITH summary_tbl AS (
    SELECT
        gender AS group_by_value
        , 'age' AS target_column
        , COUNT(*) AS row_count
        , COUNT(DISTINCT age) AS distinct_values
        , AVG(age)
        , VARIANCE(age)
        , STDDEV(age)
        , MIN(age)
        , PERCENTILE_DISC(0.25) WITHIN GROUP (ORDER BY age) AS first_quartile
        , MEDIAN(age)
        , PERCENTILE_DISC(0.75) WITHIN GROUP (ORDER BY age) AS third_quartile
        , MAX(age)
        , MODE() WITHIN GROUP (ORDER BY age)
    FROM cust
    WHERE age IS NOT NULL
    GROUP BY gender
    ORDER BY gender
), missing_tbl AS (
    SELECT
        gender AS group_by_value
        , COUNT(*) AS missing_count
    FROM cust
    WHERE age IS NULL
    GROUP BY gender
)
SELECT a.*, b.missing_count
FROM summary_tbl a LEFT JOIN missing_tbl b USING(group_by_value)
;

  (2) Apache MADlib으로 연속형 데이터의 그룹별 요약통계량 구하기

DROP TABLE IF EXISTS cust_summary;

SELECT madlib.summary('cust'     -- Source table

                      ,'cust_summary'   -- Output table

                      , 'age'                -- Target columns

                      , 'gender'            -- Grouping columns

);

SELECT column_name

FROM INFORMATION_SCHEMA.COLUMNS

    WHERE TABLE_SCHEMA = 'public'

        AND TABLE_NAME    = 'cust_summary'

    ORDER BY ORDINAL_POSITION;