Summary: 45 instances, 38 unique

Text	Count
# TODO: consider whether we need more care around stateful featurizers,	3
# TODO: Add other types of covariance estimation (e.g. Newey-West (HAC), HC2, HC3)	1
# TODO: it seems like it would be better to just return the tree itself rather than plot it;	2
# TODO: do we need to give the user more control over other arguments to fit?	2
# TODO: it seems like we need to sum over the batch because we can only apply gradient to a scalar,	1
# TODO: might be faster to break into connected components first	1
# TODO: make sure to use random seeds wherever necessary	1
rho = - moment / grad  # TODO: Watch out for division by zero!	1
# TODO: prel_model_effect could allow sample_var and freq_weight?	1
# TODO: is it right that the effective number of intruments is the	1
# TODO: consider working around relying on sklearn implementation details	2
# TODO: support freq_weight and sample_var in debiased lasso	3
# TODO: consider addding an API to DML that allows for better understanding of how the nuisance inputs are	1
# TODO: what if input is sparse? - there's no equivalent to einsum,	1
# TODO Share some logic with non-discrete version	1
# TODO: Note that there's no column metadata for the sample number - should there be?	1
# TODO: do correct adjustment for sample_var	1
# TODO: check compatibility of X and Y lengths	1
# TODO: would using einsum's paths to optimize the order of merging help?	1
# TODO: studentized bootstrap? this would be more accurate in most cases but can we avoid	1
# TODO: make sure that the public API consistently uses "T" instead of "P" for the treatment	1
# TODO: Consider investigating other performance ideas for these cases	1
# TODO: ideally, we'd also infer whether we need a GroupKFold (if groups are passed)	1
# TODO: does the numeric stability actually make any difference?	1
# TODO: any way to get this to work on batches of arbitrary size?	1
# TODO: bail out also if categorical columns, classification, random_state changed?	1
# TODO: Add a __dir__ implementation?	1
# TODO: this utility is documented but internal; reimplement?	1
# TODO: conisder working around relying on sklearn implementation details	1
# TODO: we can't currently handle unseen values of the feature column when getting the effect;	1
# TODO: should we also train a new model_y under any circumstances when warm_start is True?	1
# TODO: Check performance	1
# TODO: enrich outcome logic for multi-class classification when that is supported	1
# TODO: update docs	1
# TODO: this utility is even less public...	1
# TODO: any way to avoid creating a copy if the array was already dense?	1
# TODO: support freq_weight and sample_var	1
# TODO: is there a more robust way to do this?	1