Open MalteKurz opened 2 years ago
The actual root cause in the example above is not a not finite prediction but a propensity score estimate of 1.
The actual root cause in the example above is not a not finite prediction but a propensity score estimate of 1.
Estimated probabilities / propensity scores may need special attention, i.e., a check that they are (strictly) in the interval (0,1)
. See also: https://github.com/DoubleML/doubleml-for-py/issues/129
There is no exception handling in-place in case some learner produces infinite or missing predictions. Basically, very silently the estimates are becoming
NA
's without a warning or exception.See for example:
It is then getting even more confusing if one thereafter calls the method
bootstrap()
. This results in exceptionwhich is obviously not the root cause and also the remark to apply
fit()
will obviously not fix the issue.I propose to implement a check for finite predictions similar to the check in the Python package: https://github.com/DoubleML/doubleml-for-py/blob/b3cbdb572fce435c18ec67ca323645900fc901b5/doubleml/_utils.py#L204-L208