Please upload a sample dataset.
My initial thoughts are you should be able to identify posts, put that on columns or rows and the number of likes on the opposite column or row. I'm not sure how you would extract the most common phrase out of your data, but I'd probably take the approach of splitting each word into a new column and then doing some kind of 'most common first word, most common second word' approach.
Hi Peter, thanks for your help. I uploaded the small sample with filters. I don't know how to split post, postlike and comment as they are under the same filter section. If the postlike or comment is related to the post, they share the same ID. Maybe I could somehow do a union or something that I would filter the post with they ids and then in other file filter comments and their parentids as well and then eventually see how many ids are used per a particular post. I have the idea how it should be but I do not know how to make it work. Maybe you have any suggestions?
Is this roughly what you're looking for? Also find the workbook attached.
Topic modeling is a huge and complex field, so you may have more luck asking that question as a separate post. I've never seen phrase-based topic modeling done solely in Tableau. It's more common to see topic modeling done outside of Tableau, and the results brought into Tableau for analysis.
FacebookLikes_v10.2.twbx 17.2 KB