search

dsc-jiit-128 / ML_classification_Hacktoberfest

This repository was created for people who are new to Open-source who can learn more and be an asset to the human society :octocat: This repository will help you understand the basics of Git, GitHub and hone you Machine Learning Skills!
MIT License
1 stars 10 forks source link

Getting best train-test split value #13

Closed sagnik1511 closed 3 years ago

sagnik1511 commented 3 years ago

This PR contributes to #4.

Updates in the PR: The new train test split has been added to the notebook.

Improvements:

Previous results having train-test split as 0.7 shared the test accuracy of 0.8722. But the new notebook was executed using 0.15 test split and found the accuracy to be improved than previous which is 0.88125.

Do share if any update is needed or not :)

sarthakpant772 commented 3 years ago

hey @sagnik1511 make your changes in different file . Like copy the data u want and make a new file under submission.

sagnik1511 commented 3 years ago

So, basically, I have to create another notebook with updates and save it alongside the original one?

sarthakpant772 commented 3 years ago

yes @sagnik1511 save that in submission folder as written in readme

sagnik1511 commented 3 years ago

Do check, I've updated the PR

sagnik1511 commented 3 years ago

also, one thing, isn't this PR going to be labeled hacktoberfest accepted?