Relationship types and cardinality - median or mode?

Where this lecture talks about enhancing the traditional crow’s foot notation with information about minimum, most likely and maximum cardinality of the thing on the “many” side of the relationship, it describes the “most likely” value as the “median”. I’m not sure this is the correct term. Given the sorted array

[0, 1, 2, 2, 2, 3, 3, 4, 5, 7, 10]

The median value in this array is the one which has an equal number of elements both before and after it. As it has 11 elements, that would be the one in the sixth position, i.e. 3.

If we really want the “most likely” value, shouldn’t we take the value which occurs the most times in the array, the mode, i.e. 2?

Sorry if this is a really picky point.

I believe that the tutor was referring to the number of items/documents in the collection, not the median. In my opinion, “median” is not the right choice of word.

Min - Minimum number of documents
Most likely - Typical/expected number of documents
Max - Maximum number of documents

1 Like

Within this same lecture, I also don’t see the need of introducing a new crow’s foot notation called “one-to-zillion”. The word “many” in a crow’s foot notation covers exactly that… zillion is many.

Sorry @Opubo_19608, I have to disagree with you on this point.

I also don’t see the need of introducing a new crow’s foot notation called “one-to-zillion”. The word “many” in a crow’s foot notation covers exactly that

There’s a difference between a one to many relationship and a one to few relationship. When we design for a relational (tabular) database, we probably don’t care whether a relationship is 1-many or 1-few, but it’s important when designing a MongoDB database.

Taking some examples from the course, a mother could have had 0, 1, 2 or 3 babies (or maybe a few more), but she’s unlikely to have had 10 thousand babies. In this scenario it would make sense to embed the details of the babies in the mother’s document. We can call this a one to few relationship.

Meanwhile, a Twitter user could have followers which number anywhere between 0 and infinity. If they’ve got thousands of followers then it doesn’t make much sense to embed the followers (and all the followers’ tweets) in the document which represents the Twitter user. That’s a one to zillion relationship, where the entities on the “many” side of the relationship probably belong in their own collection.

1 Like

The foot is abstract enough to tell us the relationship. A combination of the Numeral Notation and the Crow Foot Notation is a good guide to indicate what numbers you’d expect to see, i.e. quantifying the relationship. I.e. the Numeral Notation quantifies the relationship.

To expand, “many” simply means 2 or more. If you then include a numeral notation of [1, 5M] within the ERD that helps to quantify the relationship.

What I’m saying is, I don’t see the need of changing the foot of a logical/abstract notation from three legs to five legs.

Simon,

this is a good point.
We tried to use “most likely” to keep it simple.
However, the right term is “mode” wherever we use “median”.
I have tagged the review with those change, and once re-recorded we will use “mode”.

Thanks,
Daniel.

@danielcoupal, it’s a better replacement, but I think that it may still cause confusion because mode refers to the value that appears more often as opposed to the likely number of items. My humble opinion of course! :cowboy_hat_face:

1 Like

Thinking more about it, we will keep “likely number of items”, however, remove “median” from the lesson.

1 Like

As at 3/9/2020, the lecture forbids embedding zillions in the one side and only allows referencing the one from a separate zillion collection

1 Like