This is usually due to missing columns. A column may no longer be present in an input file, may have been removed from Map Column Headers or may have been renamed by another process.
Columns could also be missing because the dataset they come from is no longer available to the pipeline (for example, unshared or archived). Check the reference dataset inputs and any error message shown within the affected stage.
If a reference dataset in a Join was not present when the pipeline was opened, open the Join stage to refresh it or re-add the datasets.
Causes can include the above, or a missing dataset used by Automap Values anywhere in the pipeline. This can appear on Trace and is often resolved by opening the Automap configuration. Check whether the reference dataset has been archived, and whether duplicate reference datasets with the same name exist or are not shared.
Archived/missing reference datasets are not listed in the stage inputs. To remove one, first unarchive it, remove it from the stage inputs, then archive it again.
Check the following:
In Aggregate, columns still highlighted in the left hand panel will not pass through the operation.
If columns are reordered in the master schema, the new order is not remembered until:
Workarounds:
Join keys may not match due to whitespace/case differences. Apply text cleansing such as TRIM, UPPER before the Join.
Duplicates on the right-hand side create a one-to-many join, returning multiple combinations. Ensure there are no duplicate join keys in the right dataset.
With transactional data, duplicates may be unavoidable. Consider using an Aggregate to compress the left dataset to the correct grain, producing a one-to-one join.
Check upstream Map Column Headers and Aggregate stages to ensure the column is mapped/included. Stages highlighted in red indicate a Trace error; resolve the error to expose column names. Also confirm the operation is turned on.
‘NaN’ indicates a non-numeric interaction (e.g. text used in a numeric expression) or a divide-by-zero. Use validations to identify non-numeric fields (e.g. ISNUMBER(<value>)).
Email support@quantemplate.com with as much detail as possible:
These details help us reproduce and resolve issues. If it’s blocking progress, we’ll try to suggest a temporary workaround while we work on a fix.
If you can’t resolve your issue, please email support@quantemplate.com with what you’re trying to do and how the actual output differs from what you expect. If you share a pipeline with us, remember to share dependent datasets from the Data repo as well.