Zulfiqar's web

Olap Cubes: Operators

Operators such as +, - , * act both as arithmetic and set. Acting as set operator

+ : creates the union of sets
For example: {set1} + {set2}
* : creates the cross product of sets.
For example: {set1} * {set2}

Happy cubing,

ZULFIQAR SYED

Olap Cube: Facts

There are 3 types of facts

• Additivie: Normally, facts are aggregated when brought in as multiple rows. We usually don't look at an individual fact record rather we look at multiple records and most of the time, facts are added.
• Semi-Additive: These facts can only be added using ONLY some of the dimensions.
• Non-Additive: These facts can not be aggregated at all.

Happy Cubing,

ZULFIQAR SYED

Olap Cube: Attributes

There would be multiple attributes members in a given dimension table such as Color in product dimension. How does the dimension reflect only the unique members upon browsing?? It has to do with the following 3 attribute properties:

• KeyColumns
• NameColumn
• ValueColumn

Thanks

Happy Cubing..

ZULFIQAR SYED

Olap Cube: Many-to-Many Relationship

Defining a Many-to-Many Relationship 

Usually dimension is related to fact table in a one-to-many relationship. For example, one customer (dimension) is related to many sales (fact). What if a single fact (product) relates to many groups and a single group is related to many products. This is known as many-to-many relationship. Product sales fact table and group table would need an intermediate table known as productgroup table on the oltp side. On the olap side, we will create a new measure group product group. Create a new dimension called group and in dimension usage, create many to many relationship between the new measure group productgroup and group dimension.

Happy Cubing,

ZULFIQAR SYED

Olap Cube: interview questions

Check out these awesome posts on olap cubes interview questions by Dave Rodabaugh.

Thanks

Happy Cubing..

ZULFIQAR SYED

Olap Cube: Fact Dimension

Normally, we create dimensions based on dimension tables related to fact table on primary-foreign key relationships. What if we want to dimension on data itself inside the fact table. For example the order number is inside the order details fact table and we would like to slice orders amount based on order number, what do we do.

In this case we create FACT DIMENSION. Check BOL for more details.

Thanks

ZULFIQAR SYED

Happy Cubing...

Olap Cube: Referenced Relationship

What if fact table does not have a direct primary-foreign key relationship with the dimension table??

You are in luck if there is an indirect relationship via an intermediate dimension. In that case first you would define an invisible, non ordered, and non optimized (key)attribute in the intermediate dimension.
This key attribute would be used in the define relationship dialogue to create the reference relationship between the fact table, intermediate dimension and the referenced dimension.

An example would be relating sales fact table to the geography dimension via customer dimension. After this relationship we could pivot the sales with respect to cities directly.

Happy cubing,

ZULFIQAR SYED

Olap Cube: Grouping Members

To automatically group attribute members, check out the following 2 properties of the attribute in dimension designer.

DiscretizationBucketCount

DiscretizationMethod.

Thanks

Happy Cubing

Zulfiqar Syed

Data mining: Code Camp feed back..

Well, I finally got the feedback for my data mining and olap talks for san diego code camp. I have listed some of the postive and negative feedback.

• Intro to data mining session
  • An excellent presentation! Speaker was very well versed on the topic and made the
    audience very much interested in the topic.
  • Good presentation. Unfortunately I should have gone to another presentation. The information was well presented but I thought I would get more SQL
  • score = excellent
• Intermediate data mining session
  • Interesting and captivating
  • Could have used a larger font for people in the back. Suggest to have a few power point slides for introductory material.
  • score = very close to excellent
    
• Into to OLAP
  
  • Good presenter, good examples.  Made me want to start investigating OLAP for my current projects.
  • Very large topic, hit the mark given time constrant.
  • score = very close to excellent

I want to thank every one who took the time to give the feedback. For next presentations, I will keep the negative comments in mind and do my best to improve the presentation.

Thanks

ZULFIQAR SYED