The unique skill of a Machine Learning model is to identify specific instances that humans would easily miss. That needle in a haystack that can make or break your business. But how can you train a machine learning model when the data points you need to identify are extremely rare?
Fortunately, this problem is not new and there are several strategies that can be employed to tackle this common scenario. From synthetic data generation techniques, algorithm and metric selection to over and under sampling, this session will cover a variety of approaches that can be implemented through common libraries and toolsets.
Attendees of this interactive, demo-led session will leave with an understanding of how to identify imbalanced classification problems and a myriad of resolution approaches to experiment with. This session would suit those with a basic knowledge of data science although all the basics will be covered in brief.