search

Nzumino / intro-data-capstone-musclehub

0 stars 0 forks source link

Wrong test #5

Open sen1 opened 6 years ago

sen1 commented 6 years ago

https://github.com/Nzumino/intro-data-capstone-musclehub/blob/ea6ae5b91f08e69bc4acd525015580da8cdd7e1a/Mhub%20NZ%20assignment/musclehub.py#L276-L295

See the instructions at the top. You are comparing group A vs group B if they have different possibility of turning into an application. So binomial test is not appropriate here. It would be appropriate if its within one group. Since the application, not application values are categorical, a chisquare test is more appropriate here.

contingency = [[200, 50], [250, 75]]
chi2_contingency(contingency)
Nzumino commented 6 years ago

I'm not sure I understand why the binomial test couldn t be used here. Both group A and B belong to the same larger group of first gym visitors.

Ok, I see. So if group A and group B had widely different numbers of visitors, such as 3000 in A and 1000 in B, for example, the binomial test would have not worked. But since they were 2000 and 2004, it still gave "correct" p-value? Is this correct?