Excellent -- thanks very much, Chris! This is exactly what I needed for a first step.
Now, do you have any suggestions for how to deal with a lot of keywords? I see that you set up the tasting note keywords by manually creating two calculated fields for each. But, my data set contains hundreds of wines whose tasting notes contain dozens or hundreds of potential keywords.
So, it seems like either: (1) I'll need to read through all the tasting notes and manually create a calculated field for each keyword therein, or (2) I'll need a way to have Tableau automatically create a calculated field for each word that appears in all the tasting notes.
Do you have any advice?
1 of 1 people found this helpful
You want to reshape your data so that keywords are identified before tableau opens it.
Otherwise you'll need to create a million calcs, depending on how many words you want to examine. (OR, perhaps, have some secondary data source with the list of words you want to examine and do something with that.)
Consider googling "keyword search tool". There are all sorts of tools out there that will serve you better. Once you have it parsed, THEN you can use tableau to viz the results.
Just an additional input, if you create a parameter as data type string Enter Text and use that in a calculated field-
if find([Notes], [Enter text]) > 0 then [Rating] end. That would make rating calculation dynamic. Attached is the workbook for reference.
Hope this helps.
text rating.twbx 9.5 KB
You could use method 5 in The Cross Join Collection:
SELECT * FROM [data$] d, [lookup$] l WHERE ' ' + d.[Notes] + ' ' LIKE '% ' + l.[Word] + '[ ,;.:!?]%'
See more in attached files.
Ps. You need to open files with legacy connector if data source is Excel, Text, or Access.
Thanks very much, Joe, Praveena, and Kettan -- a combination of all of your answers is exactly what I needed! Thanks especially to Kettan for the idea to set up wordstems and the SQL query like that.
Have a great week,
Hi Michael , did you get to finalize this tableau dashboard? I am looking to do exactly the same and it would be very useful if you could share any insight of how you finalized it.
The last time Michael logged in was Feb 2015, so it might be a better idea to start a new thread and the current users can help you