Unlocking non-traditional data for risk analytics

Rendani Mbuvha from the Wits Actuarial Science Department will present this Actuarial Sciences Seminar.

With recent advances in computing, data gathering and storage, more data has become available to the actuary than ever before. It is incumbent on the modern actuary to extract meaningful strategic value from such data.

This seminar aims to provide practical ways for incorporating non-traditional data such as text and images to enhance our understanding of actuarial and financial risks.

The seminar will begin by introducing foundational concepts and rationale behind machine learning pipelines. Details on how to extract meaningful features (rating factors) from text and images for use in predictive models will then be discussed. This will then be motivated by practical examples of using convolutional neural networks on dashcam images to detect distracted drivers and predicting stock market movements using GLMs trained on bag-of-words features.

Add event to calendar