After checking out the vid, and doing the lab, I see
y = 1 uses 
, while y = 1 uses 
Just curious, is that the standard way of plotting?
My brain just cannot process as being a “positive” class…
1 Like
Hello @dance2die,
Welcome to our community 
I am sure everyone has their own secret to make their best plots to communicate with their audience, and I have never seen any standard to that. However, I might provide you with one reason for the lab’s choice -
Sometimes, a binary dataset can be pretty imbalanced:
- healthy people VS. cancer patients
- People with good credit records VS. Defaults
- Normal transactions VS. fraud transactions
- Normal people VS. Olympics elites
- Customers who stay VS Customers who are leaving us (Churn)
- …
In these cases, we always want to “detect” something out of the population. I would not say if those “something” are good or bad or positive or negative because it is not about comparing the two classes, but to distinguish out what we want to detect.
In short, those “something” require our attention, and a sharp, red crosses can draw our attention
Besides, speaking of imbalanced dataset, there are special metrics (e.g. Jaccard score, Recall, …) for them. In sklearn, for example, those metrics always assume the minority class (that we want to detect) to carry y = 1 label. Therefore, the use of y = 1 may be considered as some kind of “standard”.
Cheers,
Raymond
2 Likes
I was trying to delink the following because in my experience, sometimes, for example, some TV shows always use “circle” as “correct answer” or “positive answer”, and not sure if you had similar experience too:
Positive is something we want to detect (e.g. Olympics elites from normal people) and we use a sharp symbol (red cross) to denote it.
1 Like
Thank you, @rmwkwok for the detailed explanation 
I will move on from the trivial matters and focus on what’s discussed about “positive” class being what we want to detect and move on 
Okay! Good luck, @dance2die!
Raymond