Categorical¶
-
class
golem.Categorical Simple categorical distribution where an uncertainty describing the probability of error, i.e. selecting the wrong category, is spread among all other available categories.
Categories will be encoded alphabetically as an ordered variable. In practice, because true categorical variables are not yet supported in sklearn, we implement this distribution as a discrete one with the first category being encoded as 0, the second as 1, et cetera, in alphabetical order.
- Parameters
categories (list) – List of categories
unc (float) – The uncertainty in the categorical choice, i.e. probability that the queried category is not the category being evaluated.
Methods
pdfProbability density function.
cdfCumulative density function.