Data Quality

Share

04/07/2017 10:21

 

 

Data Quality

 

If you have been following my journey into data, we next come to Data Quality, another loaded set of words.

A long time ago I was asked “What’s the quality of our current timetable?”. What a question to ask, I thought. To start with I simply counted how many Groups had at least one event, so we had 80% of Groups with at least one event, but what does that mean? Some Groups may only have one event, others may be missing dozens of events.

It turns out we should approach this in a very different way, and CELCAT itself has tools that can help us to get a clearer picture of a timetable’s 'quality'.

First, let’s return to my count of events by Group. What I really needed to do was to compare the timetabled hours by Group against the planned hours (the GLH).

Next, we can look at events with no Staff, no Module and no Event Category. Then we can dig deeper and compare our timetabled staff hours against their contracted hours. This I have done for customers!

Clash checking is an important part of timetabling; not my favourite pastime, but it helps ensure that a student, for example, has a timetable which is achievable (such as their not being needed in two places at once).

Of course, there is the lower level of Data Quality; for example, is a student’s enrolment correct? Does a Group have the correct GLH? Is a student’s email address a valid email address? This comes back to how we capture and store data and where. What validation can/do we have when inputting things like phone numbers or email addresses? What validation can/do we have for checking the data we already have in our systems? These are questions we should all be asking.

You can only go so far with your checks; clearly, establishing a phone number is still live, or an email address still used is a huge leap and would be too time consuming, but ensuring a phone number is all numeric would be a start.

I spend a lot of my time looking at student attendance. A register is the last element in a long chain of data and processes. The students on the register I have in my hand compared to the students in front of me is where the quality of all that data is put to the big test. Is this a register for this class? This week? Are the students all listed? Does it say the right room? Right staff?

That was my final measure for the “quality” of the timetables. How many changes did we make each day/ week/month, to the timetable and why? When I got this number to simply staff absence and the odd room change due to maintenance, I could confidently answer the question, “The quality of our current timetable? Excellent”.

 

 

Seddon Kirk

Seddon Kirk - Data Quality