Question

I’ve been given the task of normalizing this set of data:

COURSE=(CourseID, CourseName, CourseDuration, CourseFee{
    DelegateID, DelegateName, DelegateAddress, EventID, EventName, VenueID, VenueName, VenuePrice, BookingID, BookingType, BookingDate
})

The scenario is an IT company that runs short training courses at various hotels around the country, each event being hosted by one or more presenters (hence the BookingType – either Delegate or Presenter, if Presenter then no booking charge). An event is just an instance of a course running at a particular venue. VenuePrice refers to the cost for optional bed and breakfast at the hotel venue for the intervening nights

Here is what I have come up with for 3NF:

COURSE=(CourseID, CourseName, CourseDuration, CourseFee)  
DELEGATE=(DelegateID, DelegateName, DelegateAddress)  
EVENT=(EventID, VenueID*, CourseID*, EventName, EventDate)  
BOOKING=(BookingID, DelegateID*, EventID*, BookingDate, BookingType)  
VENUE=(VenueID, VenueName, VenuePrice)  

I’d like to know if this is at all accurate, and if not, perhaps a guiding hand in the right direction?

Thanks

Answer 1

On what basis do you think that is in 3NF ?

Let’s take a really simple example, Course. Where exactly if he Functional Dependency, on what key ? How can CourseName be dependent on CourseId, when CourseDuration and CourseFee are dependent on CourseName ?

Same with the rest of the tables; Event being a little more complex, has a few more errors.

You cannot normalise, or achieve 3NF, when your starting point is to stick an ID on everything that moves.

No. First normalise the data. Achieve 3NF. I can understand a CourseCode or ShortName as something the user may use to identify course, but no Id.

After that, if and only if you need to, add an Id column and the additional index.