In order to create the desired view the underlying data needs to be normalized. The following Knowledge Base article provides instructions on normalizing data. Normalizing Data for Use in Tableau | Tableau Software
In this case, I recommend replacing the columns "New Segment" and "Old Segment" with two new columns "Segment #" and "New or Old."
Tableau Desktop 9.0 has a new pivot feature when connecting to some data sources that allows the data to be normalized when connecting rather than having to edit the data source. The following Online Help article provides instructions for using the pivot feature when connecting to Data.
After pivoting the data, here is how I would create the view:
- Drag [New or Old] to the Rows shelf
- Right-click on [New or Old] in the Rows shelf and select Measure > Count
- Drag [Segment #] to the Columns shelf
- Drag another copy of [New or Old] to the Columns shelf
- Drag a third copy of [New or Old] to Color on the marks card
I attached my workbook as well so you can see what it looks like. I hope this answers your question!
comparing dimensions.twbx 57.1 KB
Thank you very much for your reply and detailed explanations.
I understand now that data needs to be normalized for such an analysis. However, I also want to highlight the changes between segments as shown below. li
With suggested normalized data structure, I don't think I can achieve such a comparison, can I? If no, what do you suggest to do in such a case?