Say that I have a large collection of words that I am comparing two features of them. For simplicity, I will give an example of plural and singular forms of the word.
My task is to calculate the percentage of the words whose plural form appear more than singular form in a corpus.
Everything is OK about calculating the percentage. But, obviously there would be some plural words would appear more than some others (e.g.
"page": 1000, the percentage being pluralized is then
99%). On the other hand, there would be some words whose percentage might be similar, but the number of times would be much less than others (e.g.
"question": 1, percentage is still
What I would like to ask is that is there any statistical function which can reflect the difference between these two cases of percentage being pluralized?