American Express Interview Question | Resampling

Question

Is it necessary to perform resampling in your dataset? How would you initiate with this process?

in progress 0
Dhruv2301 4 years 2 Answers 797 views Great Grand Master 0

Answers ( 2 )

  1. Priyabrata017
    0
    Reply

    Resampling is necessary when the dataset is imbalanced. If we have a balanced dataset, then resampling is not required. There are various ways of performing resampling:
    Oversampling : Creating copies of minority class
    Undersampling: Removing some instances of majority class
    SMOTE ( Synthetic Minority Oversampling Technique): Creating sythetic data nearest to the minority class

  2. swap007
    0
    Reply

    Generally, if you are using ensemble methods, your each model will need to be exposed to
    different kind of data. So, resampling is necessary in such cases. Also, when training data size
    is not too large, resampling is needed.

Leave an answer

Browse
Browse