The Feature class needs a label, e.g. "heights" or "firstnames".
If the feature values are numerical we may want to "bin" them to reduce the number of possible feature values. The heights from our persons have a huge range and we have only 50 measured values for our Naive Bayes classes "male" and "female". We will bin them into ranges "130 to 134", "135 to 139", "140 to 144" and so on by setting bin_width to 5.
There is no way of binning the first names, so bin_width will be set to None.
The method frequency returns the number of occurrencies for a certain feature value or a binned range.