Data architecture is not static, and you may occasionally add new data fields or modify existing structures. However, when you change the architecture of a table that’s connected to Pecan, this can cause a mismatch between the data on your side and the data that has been imported to Pecan.
Why is this a problem? When Pecan begins training a model or generating predictions, it verifies that there's a match between the relevant columns on each side. If a mismatch is detected, Pecan may be asked to rely on data that's no longer available – and the process is terminated.
A data mismatch may also make it impossible to sync data from a table that has already been imported to Pecan.
A mismatch might occur in the following cases:
A column that’s in use by Pecan has been removed from your data (the source table).
The name of the column has changed.
The “type” of the column has changed (e.g. from Integer to String).
In cases where this impedes a process, there are two potential solutions:
Modify your original table so it’s aligned with the table that was originally imported to Pecan (e.g. re-introduce a deleted column, change a column name).
Import the entire table again. You can do so by clicking the “Data” tab in Pecan, locating the relevant table, and clicking “Re-import table”.