I'm trying to get started with Tableau again and, being really new at this, there's a concept that I'm still trying to work around here.\
My use case is to pull data from an application in CSV form and build a workbook that allows me to connect to the data source and have my dashboards update automatically. Eventually, I want to connect to the database and not have to worry about the CSV. But, in the meantime, updated Dashboards!!
I get that Tableau isn't a replacement for spreadsheets since spreadsheets can be a source of data for Tableau and Tableau can use the data to create the visualizations needed from the data.
However, in spreadsheets, I can take raw data, do calculations on them and drop them into different tables or into rows in the same table. I can then use the calculated fields to create my graphs.
I do not seem to be able to do this in Tableau and, intuitively, it feels like I should be able to.
For instance, if I have a table that has two rows:
- Average
- Standard Deviation
And in the columns I have various measurements for various items.
I then add a row that does a diff (or a %average difference or...whatever). So you get:
- Average
- Standard Deviation
- Average %Error (for instance)
It seems I should be able to do this with a calculated field but, whenever I add that calculated field and drag it t the rows shelf, it adds it as a column. This makes it impossible to do the visualization I want to do. So for instance,
I calculate the Diff from Expected, the Average % Error and so forth as new rows in the table and then visualize with bar graph or line graph, etc.
However, I can't build this type of structure and then I have trouble with getting to the visualization.
Do I have to create a separate sheet with all my calculations and work from there? Do I just stick with Excel and I'm trying to use Tableau for something it's not designed for?
Rick,
Both Ajai and Norbert are correct, we really need some mock-data to be able to help you better.
Just as an example, however, if your data looked something like this:
Then, you could easily come up with some calculated fields for all of your various measures: average dimension, standard deviation of measurement, % error, difference from expected, etc. And you could create a couple of worksheets to display the information, and you might be able to get a viz that contains what you need:
If your data is not structured in that way I described above, then a different solution would be needed. As you can see, knowing the structure of your data, and being able to modify the structure, is the key to determining how to get the viz you want.
Hope this helps. Please let us know.
Kaz.
