Correct in Top N Subtab
This topic describes part of the functionality of Genesys Content Analyzer.
When a model classifies a text object, it returns a list of categories and the probability that the object belongs to them. Ranking the returned categories with the highest probability first, how likely is it that the correct category appears within the top two, the top three, and so on? Ratings of this likelihood are displayed on the Correct in Top N subtab, shown in "Models Tab: Correct in Top N Subtab."
Look at the first row for a good indication of the overall quality of the model. Its general meaning is that a single classification attempt using this model will be correct for this percentage of categories.
The other rows can help you further assess model quality. A very accurate model may have a 95 per cent probability that the correct category appears in the top two or three. For a less accurate model, you may have to go down to the top five or six to achieve 95 per cent coverage.
You can also use this rating to advise agents how many categories to look at when choosing a standard response. If there is a 95 per cent probability that the right category is in the top three, you can advise agents to examine only the top three categories.
See also Applying the Ratings.