Add source-specific checks
In order to add source specific checks, decide what the parameters will be for the check. For example, if you need to check that a currency column has a certain value, you may want to add a "currency" parameter to the collection that the source is in by going to the edit page for the collection:
You now want to add a value for this parameter to one of your sources, e.g.
Go to the drilldown view for the snapshot. You can now create a formula here to test before creating a calculated column, for example:
if(OR(ISBLANK(OUTPUT('priceCurrency')), ISBLANK(PARAM('currency'))), NULL, if(PARAM('currency')=OUTPUT('priceCurrency'), TRUE, FALSE))
This is displayed in the first column:
This formula will return a
null if we don’t have an expected value or a value, otherwise a
true value if the currency is what we expect it to be or a
false if not.
You now need to go to the schema to add a new calculated internal column that is a boolean:
We now need to import some data into a snapshot and check that the data is populated correctly. THis will also add the metric to the available metrics to set up a check on.
We now can add a manual check on the metric value for the column:
|In the future, this value will be checked statistically, although you can still add a manual check.|
pct_true check for the
booleanValue metric of the column, e.g.
This will be the percentage of the time there was a currency and an expected currency, and they were different.
|If you are always expecting a currency value, add a "required" validation also.|
You can either set it to be less than an expected value using the
VALUE check, e.g. less than 1%, or you can use the
PCT check to check it is less than a % swing from the nearest snapshot to have passed QA (excluding PUSH_IGNORE) in the last 30 days to have started collection before the snapshot being evaluated.