The numbers are estimates based on zipped size. But it is a bad idea. I should write a program to extract each zip, and count zero bytes padding at the end of the file.

Too late for today, I will write that tomorrow and update the article.