C3_W1_Anomaly_Detection - exercise 2 question

Course Q&A Machine Learning Specialization Unsupervised Learning, Recommenders, Reinforcement
1

Given that

  • 𝑝(π‘₯)<πœ€, then it is classified as an anomaly.
  • 𝑑𝑝 is the number of true positives: the ground truth label says it’s an anomaly (p=0) and our algorithm correctly classified it as an anomaly (p=0).

The hints suggest using ((predictions == 1) & (y_val == 1)).

Wondering why it isn’t ((predictions == 0) & (y_val == 0))?

2

Hello @Buddhima,

I think you had the question because in your second bullet point it mentioned p=0 two times. However, in my lab file, I do not see them:

image
image903Γ—44 2.83 KB

Therefore, to me, there might be something wrong with your lab file. If you would like to get a latest copy of the lab file, follow these instructions which will ask you to rename your current file before the system can write a new one to your account.

Without the p=0 there, and with the latest version of the lab, I believe you can clear your question by reading the lab again?

Cheers,
Raymond

PS: the lab’s section 2.3.2 told you what label value is used for representing an anomalous example.