Transformer des lignes en colonnes et inversement

Nettoyer des données dans des bases de données SQL Server

Miriam Antona

Software Engineer

Tableaux croisés dans les chiffriers

Très courant
Regrouper des données selon un ensemble précis de colonnes
Calculer des statistiques sur d'autres colonnes

Utiliser PIVOT

PIVOT : transforme les valeurs uniques d'une colonne en plusieurs colonnes.

Utiliser PIVOT – Transformer les noms de produits en colonnes

SELECT * FROM paper_shop_monthly_sales

| product_name_units | year_of_sale | month_of_sale |
|--------------------|--------------|---------------|
| notebooks-150      | 2018         | 1             |
| notebooks-200      | 2019         | 1             |
| notebooks-30       | 2019         | 2             |
| pencils-100        | 2018         | 1             |
| pencils-50         | 2018         | 2             |
| pencils-130        | 2019         | 1             |
| crayons-80         | 2018         | 1             |
| ...                | ...          | ...           |

Utiliser PIVOT – Transformer les noms de produits en colonnes

Passer de

| product_name_units | year_of_sale | month_of_sale |
|--------------------|--------------|---------------|
| notebooks-150      | 2018         | 1             |
| notebooks-200      | 2019         | 1             |
| pencils-50         | 2018         | 2             |
| crayons-80         | 2018         | 1             |
| ...                | ...          | ...           |

| year_of_sale | notebooks | pencils | crayons |
|--------------|-----------|---------|---------|
| 2018         |  150      |  150    |  80     |
| 2019         |  230      |  130    |  170    |

Utiliser PIVOT – Transformer les noms de produits en colonnes

SELECT
    year_of_sale,
    notebooks,
    pencils,
    crayons
FROM
    (SELECT
    year_of_sale,
    SUBSTRING(product_name_units, 1, charindex('-', product_name_units)-1) AS product_name,
    CAST(SUBSTRING(product_name_units,
        charindex('-', product_name_units)+1, len(product_name_units)) AS INT) units
    FROM paper_shop_monthly_sales) AS sales
PIVOT (SUM(units)
FOR product_name IN (notebook, pencils, crayons))
AS paper_shop_pivot

Utiliser PIVOT – Transformer les noms de produits en colonnes

SELECT
    year_of_sale,
    notebooks,
    pencils,
    crayons

Utiliser PIVOT – Transformer les noms de produits en colonnes

SELECT
    year_of_sale,
    notebooks,
    pencils,
    crayons
FROM
    (SELECT
    year_of_sale,
    SUBSTRING(product_name_units, 1, charindex('-', product_name_units)-1) AS product_name,
    CAST(SUBSTRING(product_name_units,
        charindex('-', product_name_units)+1, len(product_name_units)) AS INT) units
    FROM paper_shop_monthly_sales) AS sales

Utiliser PIVOT – Transformer les noms de produits en colonnes

| year_of_sale | product_name | units |
|--------------|--------------|-------|
| 2018         | notebooks    | 150   |
| 2019         | notebooks    | 200   |
| 2019         | notebooks    | 30    |
| 2018         | pencils      | 100   |
| 2018         | pencils      | 50    |
| 2019         | pencils      | 130   |
| 2018         | crayons      | 80    |
| 2019         | crayons      | 90    |
| 2019         | crayons      | 80    |

Utiliser PIVOT – Transformer les noms de produits en colonnes

SELECT
    year_of_sale,
    notebooks,
    pencils,
    crayons
FROM
    (SELECT
    year_of_sale,
    SUBSTRING(product_name_units, 1, charindex('-', product_name_units)-1) AS product_name,
    CAST(SUBSTRING(product_name_units,
        charindex('-', product_name_units)+1, len(product_name_units)) AS INT) units
    FROM paper_shop_monthly_sales) AS sales
PIVOT (SUM(units)

Utiliser PIVOT – Transformer les noms de produits en colonnes

SELECT
    year_of_sale,
    notebooks,
    pencils,
    crayons
FROM
    (SELECT
    year_of_sale,
    SUBSTRING(product_name_units, 1, charindex('-', product_name_units)-1) AS product_name,
    CAST(SUBSTRING(product_name_units,
        charindex('-', product_name_units)+1, len(product_name_units)) AS INT) units
    FROM paper_shop_monthly_sales) AS sales
PIVOT (SUM(units)
FOR product_name IN (notebook, pencils, crayons))

Utiliser PIVOT – Transformer les noms de produits en colonnes

SELECT
    year_of_sale,
    notebooks,
    pencils,
    crayons
FROM
    (SELECT
    year_of_sale,
    SUBSTRING(product_name_units, 1, charindex('-', product_name_units)-1) AS product_name,
    CAST(SUBSTRING(product_name_units,
        charindex('-', product_name_units)+1, len(product_name_units)) AS INT) units
    FROM paper_shop_monthly_sales) AS sales
PIVOT (SUM(units)
FOR product_name IN (notebook, pencils, crayons))
AS paper_shop_pivot

Utiliser PIVOT – Transformer les noms de produits en colonnes

| year_of_sale | notebooks_units | pencils_units | crayons_units |
|--------------|-----------------|---------------|---------------|
| 2018         |  150            |  150          |  80           |
| 2019         |  230            |  130          |  170          |

Utiliser UNPIVOT

UNPIVOT : transforme des colonnes en lignes.

SELECT * FROM pivot_sales

| year_of_sale | notebooks | pencils | crayons |
|--------------|-----------|---------|---------|
| 2018         |  150      |  150    |  80     |
| 2019         |  230      |  130    |  170    |

Utiliser UNPIVOT – Transformer les noms de produits en lignes

SELECT * FROM pivot_sales
UNPIVOT
    (units FOR product_name IN (notebooks, pencils, crayons)
) AS unpvt

| year_of_sale | units | product_name |
|--------------|-------|--------------|
| 2018         | 150   | notebooks    |
| 2018         | 150   | pencils      |
| 2018         | 80    | crayons      |
| 2019         | 230   | notebooks    |
| 2019         | 130   | pencils      |
| 2019         | 170   | crayons      |

Passons à la pratique !

Nettoyer des données dans des bases de données SQL Server