Data Warehouse Dimensional Modeling – Alphabetic Grades?
2 posters
Page 1 of 1
Where do alphabetic grades belong?
Data Warehouse Dimensional Modeling – Alphabetic Grades?
Hello!
First off, thank you for taking time to read this post. As a data modeler on a BI team in an higher education setting, I am struggling with a certain dimensional modeling dilemma and am not sure how to handle it. The dilemma I am struggling with is what to do with grades - alphabetic grades(A,B,C,D,F,WX, etc). We closely follow Kimball's approach for data modeling - one model for each business process. Poking around on Kimball's page, I found an article that mentions putting grades on a fact table and not in a dimension table(http://www.kimballgroup.com/2003/01/fact-tables-and-dimension-tables/). However, he doesn't say if the grade is numeric(60, 70, 80, 90 ,100, etc) or if it is alphabetic like mentioned above. The article makes it seem like it belongs directly on the fact table. I can understand that if a grade is numeric, it belongs on the fact for sure - it's a true numeric measure. But if the grade is alphabetic only, do we have a dimension table that holds all of those available grades that links to the fact table through a sequential ID number? Do we put them directly on the fact table? Maybe they belong in both places? Maybe it doesn't matter at all? There are factless facts and this could be a scenario for one but I wanted to check and see how other institutions in higher education have handled this dilemma. I have not yet encountered a dilemma like this in my data modeling career thus far and cannot foresee the negative and positive impacts creating the model one way or another may have in the long term. How has your intuition handled this scenario in the context of data modeling?
Thank you for your time and input!
First off, thank you for taking time to read this post. As a data modeler on a BI team in an higher education setting, I am struggling with a certain dimensional modeling dilemma and am not sure how to handle it. The dilemma I am struggling with is what to do with grades - alphabetic grades(A,B,C,D,F,WX, etc). We closely follow Kimball's approach for data modeling - one model for each business process. Poking around on Kimball's page, I found an article that mentions putting grades on a fact table and not in a dimension table(http://www.kimballgroup.com/2003/01/fact-tables-and-dimension-tables/). However, he doesn't say if the grade is numeric(60, 70, 80, 90 ,100, etc) or if it is alphabetic like mentioned above. The article makes it seem like it belongs directly on the fact table. I can understand that if a grade is numeric, it belongs on the fact for sure - it's a true numeric measure. But if the grade is alphabetic only, do we have a dimension table that holds all of those available grades that links to the fact table through a sequential ID number? Do we put them directly on the fact table? Maybe they belong in both places? Maybe it doesn't matter at all? There are factless facts and this could be a scenario for one but I wanted to check and see how other institutions in higher education have handled this dilemma. I have not yet encountered a dilemma like this in my data modeling career thus far and cannot foresee the negative and positive impacts creating the model one way or another may have in the long term. How has your intuition handled this scenario in the context of data modeling?
Thank you for your time and input!
bertalot17- Posts : 5
Join date : 2013-03-26
Re: Data Warehouse Dimensional Modeling – Alphabetic Grades?
It's a dimension, but does it need a dimension table?
If there are other attributes associated with the letter grade, then you would typically have a dimension table to hold those attributes. If all there is is the letter grade itself, then you store it as a degenerate dimension in the fact.
However, it is common that a grade also has a numerical score in order to calculate GPA and such. If the rules for assigning the score are fixed, there is no reason you cannot store both the letter grade and score on the fact. The score is a measure.
If there are other attributes associated with the letter grade, then you would typically have a dimension table to hold those attributes. If all there is is the letter grade itself, then you store it as a degenerate dimension in the fact.
However, it is common that a grade also has a numerical score in order to calculate GPA and such. If the rules for assigning the score are fixed, there is no reason you cannot store both the letter grade and score on the fact. The score is a measure.
Re: Data Warehouse Dimensional Modeling – Alphabetic Grades?
Thank you. I will look into this method a bit more.
bertalot17- Posts : 5
Join date : 2013-03-26
Similar topics
» Using the Dimensional Data Warehouse as source data for the OLTP process
» Data Warehouse Modeling
» Is it the end of the Relational Dimensional Data Warehouse ?
» Data Modeling question (not really dimensional though)
» effect of data sources in dimensional modeling
» Data Warehouse Modeling
» Is it the end of the Relational Dimensional Data Warehouse ?
» Data Modeling question (not really dimensional though)
» effect of data sources in dimensional modeling
Page 1 of 1
Permissions in this forum:
You cannot reply to topics in this forum