me, myself & andrei

13 Nov 2008
Data Warehousing Comment

I have a post underway regarding Period 1 at KTH, but before I get to that, I’ll post one of my messages regarding a Data Warehouse assignment and the way it has been handled.

I think I speak in the name of many students. We’ll see..

After today’s seminar, I am even more confused about building DW schemas.

  • Example -
    Excerpt from Assignment 1 Description: “Traditionally the bank has had an account oriented approach and it is now felt that they would like to assume a more customer oriented approach in their analysis.”

Seminar: During the second seminar, at some point, it was highlighted that at transactional level you cannot connect the Customer Dimension in any way, simply because of a lack of linkage in the source-data between customer and transaction. Soo all transactional designs having a Customer Dimension are not good. Period.

1. So basically they want a customer-oriented approach, but we, as DW designers, just tell them that it is not possible?! Isn’t there supposed to be some sort of cohesion between the source-data and the data warehouse? It’s ok to tell them it’s not possible now, but we should advise for source-data handling changes as well (I think).

  1. By chance I have been present at two seminars, and many (all ?!) groups have gone with a Customer Dimension. Now, is that because we just don’t understand DW? Is that because the assignment’s description wasn’t clear enough? Or maybe both? Whatever it is, the same way we, as students, need to question our knowledge and improve our skills, the DW explanations should also adjust. Otherwise, the only conclusion that one can draw is that EMIS right now, doesn’t have good enough students.

  2. Seminars should allow each group to present their own work all the way. I cannot relate my work to several other projects, that I didn’t read thoroughly. Presenting one part of a project for less than 10 minutes (both ways - students and seminar leader, while going through description, bad and good points, and not highlighting the important aspects and their reasoning) is simply filling in time by seeing lots of schemas, but no “best implementation” of schema X.

    • End of the example -

Along with the above example, there were other small things that are in the same spirit.

Even if there’s no “best implementation” scenario in a seminar, how come there’s no exemplification or “a possible answer to the assignment”? As I have seen during the 2nd seminar, there is one, but apparently it cannot be published. Two things need to be highlighted here:

a. Is this assignment-description so valuable, and so unique that another similar assignment (that can have a possible answer ready to be published) cannot be developed to check for more or less the same requirements? Is it that no matter if you cannot publish “one possible solution”, you need to stick to this assignment taken from a real-life scenario? I’m finding this hard to believe.

b. Since the students in the 2nd seminar were shown the schemas built for the assignment (despite not having the right to publish them, apparently), I find this on the border-line of discrimination. Some 40 people now have a good idea about how they should have handled the assignment, some other 40 people don’t.

Finally, I think the basic foundation of education is not teaching what rules to follow, but modeling your knowledge and understanding in order to follow certain rules. Today we shared bits and pieces of our work, without clearly defining what is bad or what is good (with the above example as an exception - and that only happened during one seminar only). I have no doubt that it is hard to get into the mind of a group of students, to see their reasoning and to show the flaws, but ultimately this was a true gain during our ESM course, period 1. We, as determined students, don’t want just to pass EMIS courses, or EMIS course assignments. I’m sure that while many or all of us want good grades, ultimately we want to gain practical knowledge and to be able to list our improvements without looking at the course goals’ list. And that cannot be achieved by simply sharing our work among each other, nor by being told this is wrong. Maybe my opinion is subjective.

Just for clarification, I’m not arguing that I or any of my colleagues are exceptional students (although we should be, after what-should-have-been a harsh admission to a prestigious university and a popular programme). Nor am I complaining and whining. But I think and ultimately hope that it’s valuable for you to know the above.

Coming back to present-time, so with no possibility of having the real-life answer that the DW designer had to this real-life assignment, and only with a future seminar of some sort with a couple of groups presenting their A1 projects as the best there were and explaning their reasoning, is it possible to have at least a list of guidelines that we were supposed to follow?
1. You should have excluded the Customer Dimension from the transactional level schemas due to..
2. You should have had socio-demographics as mini-dimension and not as an outrigger, in order to..
3. etc, etc.

Med vänliga hälsningar,

PS: if you, as a student, feel that you stand by most of the above, it would be nice if you reply with a simple “I support this” in order to show that I am not rambling on my own