Two predictions are co-occurring if they're predicted on the same feedback record.

For example, in the attached image, Email , Meeting and Issue with Email are co-occurring predictions.

While Quantifying feedback, if you've added Tracked Keywords or Reasons in your filter, the results by default include co-occurring predictions, along with the filtered predictions.

For example, in the attached Quantify with a query to show Tracked Keywords and with a filter for the Tracked Keyword Email, the results include Tracked Keywords that occur along with the Tracked Keyword Email on feedback records.

Looking at co-occurring predictions can be a great way to find adjacent topics that customers discuss when they talk about your feature area. You can identify whether a prediction in your results is co-occurring, or whether it's directly related to the applied filters, by the chain-link icon next to co-occurring predictions.

You can turn-off co occurring predictions by toggling the control on the top-right of your Quantify table.

Controlling the co-occurring toggle can be even more useful while Quantifying Reasons with Tracked Keywords in the filter, or vice versa. Here as well, by default, results include co-occurring predictions.

Upon toggling off co-occurring from the top-right of your Quantify table, results include predictions that are either in your filter, or are "related" to predictions in your filter. Here, "related" implies a relation between a Tracked Keyword and a Reason.

A Reason is related to a Tracked Keyword if it talks about it. We determine whether a Reason talks about a Tracked Keyword by checking whether the Reason mentions the Tracked Keyword, or any of its phrases. These relations, once established, are maintained across reason merges. When you modify the definition of a Tracked Keyword or create new Tracked Keywords these relations dynamically evolve.