Whenever you perform arithmetic across SQL columns, take care to handle NULL values appropriately.
With the unveiling of Fivetran Transformations, you now have even more reason to perform arithmetic operations across columns within your data warehouse. Unfortunately, a case may arise where a value is missing, but its absence should not invalidate an entire record.
Suppose you have sales records from a variety of jurisdictions and product categories. Not all jurisdictions have sales taxes and some product categories come with mandatory hazardous materials fees:
item | price | tax | hazmat_fee |
---|---|---|---|
oat_bran | 1.00 | 0.06 | NULL |
scalpel | 5.50 | NULL | NULL |
mercury_fulminate | 200.00 | NULL | 35.00 |
cesium_iodide | 450.00 | 45.00 | 35.00 |
If you tried to calculate the total cost for each record using the following query:
SELECT
item,
price,
tax,
hazmat_fee,
price + tax + hazmat_fee as total_cost
FROM
sales.records
Your results will look like so:
item | price | tax | hazmat_fee | total_cost |
---|---|---|---|---|
oat_bran | 1.00 | 0.06 | NULL | NULL |
scalpel | 5.50 | NULL | NULL | NULL |
mercury_fulminate | 200.00 | NULL | 35.00 | NULL |
cesium_iodide | 450.00 | 45.00 | 35.00 | 530.00 |
What you really want in this case is to default to a value of 0 whenever a NULL
is encountered. You can do this using the COALESCE function, which takes the first non-NULL argument from a list):
SELECT
item,
price,
tax,
hazmat_fee,
price + COALESCE(tax, 0) + COALESCE(hazmat_fee, 0) as total_cost
FROM
sales.records
Returning:
item | price | tax | hazmat_fee | total_cost |
---|---|---|---|---|
oat_bran | 1.00 | 0.06 | NULL | 1.06 |
scalpel | 5.50 | NULL | NULL | 5.50 |
mercury_fulminate | 200.00 | NULL | 35.00 | 235.00 |
cesium_iodide | 450.00 | 45.00 | 35.00 | 530.00 |
Remember - when adding across columns, one plus NULL equals NULL! This is true for subtraction, multiplication, and division, too.
Learn more about Fivetran Transformations with a personalized demo or a free trial.
Launch any Fivetran connector instantly.