Open jonkeane opened 6 years ago
If a user tries to send a known-malformed set of weights we should try and warn gently.
For instance: if a category that has 0 observations is given a weight, raking will never converge.
On the note of improvements to makeWeight, I think missing categories should not need to be included in the targets even if the missing category is in the middle of the non-missing categories.
If a user tries to send a known-malformed set of weights we should try and warn gently.
For instance: if a category that has 0 observations is given a weight, raking will never converge.