Implement the changes required for the lasso regression ML thing. Everything should be working, but here is what I have verified:
Full run (--all) completes without error
Generates reasonable bootstrap replicates
Computes the Correct log likelihood
I think this should be sufficient for the researchers in Israel to finish their work, but please let me know if more is required.
A quick note about the FloatWeightVector. I decided that this was needed for 2 reasons:
Floats are sufficiently different from integers to require a distinction in weights.
WeightVector was used in contexts were using floats was inappropriate.
But I can understand if you want only one WeightVector type. If this is the case, I will compact them into a single type by migrating the old WeightVectors to something like std::vector<uintType>
Implement the changes required for the lasso regression ML thing. Everything should be working, but here is what I have verified:
--all
) completes without errorI think this should be sufficient for the researchers in Israel to finish their work, but please let me know if more is required.
A quick note about the
FloatWeightVector
. I decided that this was needed for 2 reasons:WeightVector
was used in contexts were using floats was inappropriate.But I can understand if you want only one
WeightVector
type. If this is the case, I will compact them into a single type by migrating the oldWeightVectors
to something likestd::vector<uintType>