You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The skew computation and KL divergence metrics accept an alpha parameter to add Laplace smoothing. However, this parameter is not exposed via the higher level APIs. We should be able to supply a custom value of smoothing to apply, since the default of add one smoothing is not always appropriate for a given distribution (eg: add one smoothing for a probability distribution skews it more towards a uniform distribution, while the same smoothing is okay when we're dealing with a distribution of counts).
The text was updated successfully, but these errors were encountered:
The skew computation and KL divergence metrics accept an
alpha
parameter to add Laplace smoothing. However, this parameter is not exposed via the higher level APIs. We should be able to supply a custom value of smoothing to apply, since the default of add one smoothing is not always appropriate for a given distribution (eg: add one smoothing for a probability distribution skews it more towards a uniform distribution, while the same smoothing is okay when we're dealing with a distribution of counts).The text was updated successfully, but these errors were encountered: