The Data Sample feature in Tableau Prep Builder leads to a discrepancy between the preview data and the actual output results.

Published: 26 Jan 2024
Last Modified Date: 02 Feb 2024

Question

The rows/data that were displayed on Cleaning step of Tableau Prep Builder disappear when they are actually output from the flow.
The People data/table shown in the Superstore example below is intentionally set the "Number of rows" to 2 in Data Sample setting in Tableau Prep. This is an issue that occurs when the data is large enough to be "sampled"

User-added image

Orders : Having all Region in real.
People : Sampled only East and West rows from Region field (having all Region in real).

In this case, only rows that Region field in the Orders data is Central as a result of the left join are passed to the next cleaning step.
User-added image

However, If user run the flow, the output row will be 0 records because all value in Region field are matched.

Environment

Tableau Prep Builder

Answer

Set the Data Sample to "Maximum" if possible (the default setting is "Automatic"), or create a Flow that ensures that the expected results match the actual output from Output Step. This is an issue with the data preview that is inconsistent with the actual output. The output is correct. The preview data before the output in Clean/Join steps are influenced and may not always show the record that will be found in the output when the data is "sampled." This behavior is by design.

User-added image