The function _infer_datetime_format() in woodwork/utils.py utilizes pd.core.tools.datetimes.guess_datetime_format() internally, which requires the input to be a datetime string. However, there is no casting the input as a string and an error is thrown if the input is not a string (even if the input is castable).
The function
_infer_datetime_format()
inwoodwork/utils.py
utilizespd.core.tools.datetimes.guess_datetime_format()
internally, which requires the input to be a datetime string. However, there is no casting the input as a string and an error is thrown if the input is not a string (even if the input is castable).This issue needs to be resolved as part of EvalML's pandas 2 update compatibility (https://github.com/alteryx/evalml/pull/4216) and is related to #1688