Question: what patterns do you normally see in large sets of data?

Sreejita Ghosh answered on 23 Jan 2020:

It depends on the kind of dataset I have and they problem I am trying to investigate in that dataset. The medical team I collaborate with are trying to understand rare diseases and hence we have less number of subjects from the ‘sick’ population. I use a technique called oversampling to handle this imbalanced class issue.
The relative amount of one hormone wrt the rest of the hormones extracted from the urine of a subject, for all the extracted hormones form profiles. Profiles of subjects suffering from the same conditions are expected to be similar. And my classifier even makes a prototype profile for each condition which helps the doctors study teh condition better.