Hi Norah

Import SMOTE here: from imblearn.over_sampling import SMOTE

Read about SMOTE in imbalanced-learn API, here: https://imbalanced-learn.readthedocs.io/en/stable/api.html

Moreover, read pages 3 and 4 from this paper:Article Optimizing Stochastic Gradient Descent in Text Classificatio...

If you need more help please let me know.

Good Luck.

Shadi

This link has a coding example which might help you. Please read the documentation provided by Shadi Diab in his answer too.

https://www.kaggle.com/qianchao/smote-with-imbalance-data

Hello Norah!

Hope you like this

https://campus.datacamp.com/courses/fraud-detection-in-python/introduction-and-preparing-your-data?ex=6

Azhar Imran Saikiranvarma Sirigiri Manoj Kumar

Hello everyone, thank you so much for your responses.

Actually, I managed to implement SMOTE and ADASYN to my classifier, but, each time I run the code I get different results for the accuracy in the same dataset is that normal?

I'm quit bit new to Machine learning this is my first project in this filed.

You need to fix the seed number so that you can replicate the result each time you perform the task.

HTH.

Dr. Samer Sarsam

Hello, Dr. Samer Sarsam

Thank you for your advice, I will try to do that and I hope it will work for me.

Norah Suhail you should fix the seed number:

import random

random.seed(7)

Azhar Imran

Thank you, It worked for me.

The previous answers are great. In terms of a theoretical context, I would suggest encoding the sentences into numeric values, which can then be possibly treated as multiple classes, and once the balancing has been done (since SMOTE only works in theory on numeric data), the encoded sentences can be inverse-transformed into their original string representations. Though I am not pretty sure if this would work, might be one aspect of applying it and giving it a try. Hope this helps.