Unlock a world of possibilities! Login now and discover the exclusive benefits awaiting you.
if interviewer asked me what challenges u faced during developing data model?what are the challenges u faced guys pls tell me i am new to qlikview?
i am greatful to u people?
Hi manojqlik,
the main challenges in building a datamodel, in my eyes, are
1) finding out where which pieces of data are stored
2) finding out how to link them
If your database, unlike ours, is well documented, neither of these should be a big issue - then I really don't think that developing a datamodel should be a great challenge. Sure, there are some things to be aware of like the automatic linking of tables which might force you to rename fields a nr. of times, but that is a minor one.
I'm sure there also are some documents or blogs on this around here. There are some "best practices" that you can follow, of course.
HTH
Best regards,
DataNibbler
P.S.: I also have some experience from a secondary database of ours - that was quite well documented, so there was nothing much to it. Some issues will invariably pop up because of the way data is populated in the database, but those don't concern the development of the datamodel.
Important points we need to consider for data model is:
1. the way the data flow
2. not required complete data from the table, we can use left keep/join to load only relevant data
3. based on the required metrics data model need to be done
4. once the data model was done we need to validate the same as per the metric logic in the db and must to do data comparison
5. join between the tables
6. decreasing the volume of the dashboard will be completely depends up on the data model
7. must avoid circular loops.
.
Are you asking us to provide you answers for interview questions???
I recently developed a model.
First I had to understand the business logic. What was I going to eventually report on? The model would reflect this logic. I typically try to use a star schema topology to model this. I also try to minimize the number of tables that I build.
Second, I want to determine the dimensions that would eventually be used for filters. This is the easier part.
Third, I try to build a fact table or transaction based table that will hold the majority of the data I am reporting on.
Fourth, I remove unnecessary fields. Consider building mapping tables.
Fifth, I look for a few major keys to tie the model all together.
Sixth, when the above is complete, I decide how to link them all together, removing synthetic keys, circular references.
Finally, test the reload and see where it is not optimized.
This is one of the best presentations I've seen on the subject.
Good luck!