Dealing with imbalanced datasets in Machine Learning

About this webcast

Live on Wednesday June 17th, 2020 at 9:00 AM – 10:00 AM PST

Get a free IBM Cloud account to get your data science projects started.

In this webcast, we will looks at a common issue for classification models: imbalanced datasets, and look at a few different techniques that can give us a starting point to deal with this nuance. An imbalanced datasets is one where the labels aren’t distributed similarly, that is, you have many more of one label than others. In these situations, the trained classifier often becomes biased towards the majority label class and tends to miss the minority class. This will be an issue in cases where we want to detect that minority class, say in fraud detection, or early detection of cancer cells.

Speakers Bio

Omid Meh is a software engineer and data scientist at the IBM Developer Advocacy team with 5+ years of experience in software development and data science. He has vast knowledge, from electronics, to full-stack web dev, to data science and enjoys solving problems of all kinds. Moreover, he loves helping others find the right solutions, so if you have a problem, you know who to reach out to for pointers!

About this series

Visit the IBM Developer Webcast Wednesdays show page for more tech talks and sign up for the Developer Webcast newsletter to get notifications for upcoming tech talks.