The frequency role in an analysis ("Freq") is not meant in the sense of time: daily, weekly, monthly, annually. Instead, it is meant in the sense of number of occurrences. It expects a count of the number of rows with the same observation levels. It is not a requirement to use this role. It is a convenience.
Here is an example with each observation of protein on its own row. (Row equals observation)
I can examine the distribution of Protein with this launch dialog without the Freq role.
I get this platform.
But sometimes the data is available as a summary.
I could manually expand this short data table (5 rows) so that each type of protein has the correct number of rows in a tall table (100 rows), but the Freq role does that for me internally.
The result is the same because the data is the same.
So the Freq role is a convenience, a way to let JMP know that this single row represents one or more observations. It is not frequency in the sense of time. It is not a weight. (Each observation still has a weight of 1.)