Hey
I deal with survey data, I have for a long time and recently decided to see how BI can develop my work.
I'm pretty new to it and I'm not sure of the esoteric data management terminologies so I'm just going to get straight to it and hopefully we can find a way to communicate!
Let's say we have a survey with 3 multiple-choice questions. Each question has 3 possible answers and only one answer can be given per question. When you pull that data you will very easily be able to organise the data in the fashion that PBI needs it to be in in order to read/use it. From there is it very simple to create your charts and explore your data. You will, for example, be able to click on a segment of the piechart you made for the first question, and that will automatically show you how those people from that segment answered the rest of the survey questions. Very useful, very easy, very good.
Now let's say we've added a multiple-choice matrix question. Let's say we're now using checkboxes instead of radio buttons also, to really spice it up. If you've followed me to this point and you know what I'm talking about, you'll know that some fairly serious cleansing is required in order to use the data you receive from pulling that survey data. Typically you'll need to unpivot the data for that question. You do this, and you end up with a 'flattened' table of the format that is required by PBI. (Can this flattened question data be merged with the rather more simple data from the other simple questions? Can anyone point me in a direction?)
And here is my problem;
I am trying to find a way to use this flattend data from various questions with the simple data from simpler questions so that I can use the features of PBI that I initially mentioned in paragraph 2.
To my knowledge, there is no way to aggregate all of this data into one table which can create charts from all the data you collect from the survey. I do not know of a way to organise data from simple questions with flattened data from more complex questions in order to do everything from one data source/table.
I have tried breaking down each question into it's own table, using the respondentID as the link between all tables (so PBI knows that answers from table x are linked to answers from table y, etc). Like an idiot I thought that creating a relationship between all these tables using the respondentID as the link would work - but apparently I cannot create a relationship because the columns need unique entries - the whole point was that they were not unique entries and that's why a relationship should have been easy to create - clearly not.
Ultimately, what I am currently trying to learn is how to take survey data formatted in excel and use that data in PBI - essentially it's a data modelling problem. I have discovered that if you have a sales list with sales dates, prices, brands etc - then PBI is just about the easiest thing in the world to use - if you're trying to display survey data, it seems to be a whole different sort of trial.