FuzzyWuzzy

What do you like best?

This Python package takes string comparison to another level by providing a similarity score (called a "ratio" in the package) when comparing strings. Rather than spitting out the standard true/false output when comparing strings for similarity, FW provides a similarity score out of 100, making it easy to find *almost* matches. This is not possible using other string comparisons like regex, simple compare, and/or difflib.

What do you dislike?

FW's methods use the Levenshtein Distance to calculate similarity between strings. Sometimes, the ideal string response (e.g. if using it for grading written responses for accuracy) is a set length, and all responses of other lengths are incorrect. In that case, a distance that's computationally simpler -- like the Hamming distance, for instance -- could be used, and this would speed up computation time, especially for a large database of responses. Hence, having the ability to switch between distance metrics would be a nice addition to FuzzyWuzzy.

What problems are you solving with the product? What benefits have you realized?

Extremely useful for quickly scoring written response answers (e.g. surveys, etc.). If a question has a correct response that can be phrased multiple ways, FuzzyWuzzy is used to compare given answers to the "ideal" response and grades can be given based on the ratio/score provided by FuzzyWuzzy's simple "ratio" method.