Funções de ranking

Análise de Séries Temporais no PostgreSQL

Jasmin Ludolf

Content Developer, DataCamp

ROW_NUMBER gera numeração sequencial

SELECT
    station_id,
    year_month,
    t_monthly_min,
    ROW_NUMBER() OVER (
      PARTITION BY station_id
      ORDER BY t_monthly_min) AS rank
FROM temperatures_monthly AS tm
WHERE year_month BETWEEN '2018-01-01'
    AND '2018-12-31'
ORDER BY station_id, t_monthly_min;

|station_id|year_month|t_monthly_min|rank|
|----------|----------|-------------|----|
|         1|2018-12-01|          5.4|   1|
|         1|2018-02-01|          6.2|   2|
|         1|2018-01-01|          7.4|   3|
|         1|2018-11-01|          9.0|   4|
|         1|2018-03-01|         10.6|   5| 
|         1|2018-04-01|         14.7|   6|
|         1|2018-10-01|         16.6|   7|
|         1|2018-05-01|         17.2|   8|
|         1|2018-06-01|         21.9|   9|
|         1|2018-09-01|         24.4|  10|
|         1|2018-07-01|         28.1|  11|
...

Rank

RANK(): atribui uma classificação a cada linha em cada partição conforme o ORDER BY, começando em 1; repete a classificação para valores iguais

...
RANK() OVER (
        PARTITION BY station_id 
        ORDER BY t_monthly_min DESC) AS rank
...

...
ROW_NUMBER() OVER (
        PARTITION BY station_id 
        ORDER BY t_monthly_min DESC) AS row
...

Rank vs. row

RANK()

|year_month|t_monthly_min|rank|
|----------|-------------|----|
|2018-02-01|          1.6|   1|
|2018-01-01|          1.6|   1|
|2018-03-01|            2|   3|
|2018-11-01|          2.5|   4|
|2018-12-01|          2.7|   5|
|2018-04-01|          3.4|   6|
|2018-10-01|          4.1|   7|
|2018-09-01|          4.7|   8|
...

ROW_NUMBER()

|year_month|t_monthly_min|rank|
|----------|-------------|----|
|2018-02-01|          1.6|   1|
|2018-01-01|          1.6|   2|
|2018-03-01|            2|   3|
|2018-11-01|          2.5|   4|
|2018-12-01|          2.7|   5|
|2018-04-01|          3.4|   6|
|2018-10-01|          4.1|   7|
|2018-09-01|          4.7|   8|
...

Dense rank

DENSE_RANK(): atribui uma classificação a cada linha em cada partição conforme o ORDER BY, começando em 1; repete a classificação para valores iguais
- Não pula classificações

...
DENSE_RANK() OVER (
    PARTITION BY station_id 
    ORDER BY t_monthly_min DESC) AS rank
...

Rank vs. dense rank

RANK()

|year_month|t_monthly_min|rank|
|----------|-------------|----|
|2018-02-01|          1.6|   1|
|2018-01-01|          1.6|   1|
|2018-03-01|            2|   3|
|2018-11-01|          2.5|   4|
|2018-12-01|          2.7|   5|
|2018-04-01|          3.4|   6|
|2018-10-01|          4.1|   7|
|2018-09-01|          4.7|   8|
...

DENSE_RANK()

|year_month|t_monthly_min|rank|
|----------|-------------|----|
|2018-02-01|          1.6|   1|
|2018-01-01|          1.6|   1|
|2018-03-01|            2|   2|
|2018-11-01|          2.5|   3|
|2018-12-01|          2.7|   4|
|2018-04-01|          3.4|   5|
|2018-10-01|          4.1|   6|
|2018-09-01|          4.7|   7|
...

Percent rank

PERCENT_RANK: atribui uma classificação a cada linha em cada partição conforme o ORDER BY, em percentual
- (rank - 1) / (total de linhas da partição - 1)
- Valores float de 0 a 1

...
PERCENT_RANK() OVER (
    PARTITION BY station_id 
    ORDER BY t_monthly_min DESC) AS percent_rank
...

Saída do percent rank

|year_month|t_monthly_min|percent_rank|
|----------|-------------|------------|
|2018-02-01|          1.6|           0|
|2018-01-01|          1.6|           0|
|2018-03-01|            2|        0.18|
|2018-11-01|          2.5|        0.27|
|2018-12-01|          2.7|        0.36|
|2018-04-01|          3.4|        0.45|
|2018-10-01|          4.1|        0.54|
|2018-09-01|          4.7|        0.63|
...

Vamos praticar!

Análise de Séries Temporais no PostgreSQL