Empowered by latest development in machine learning and big data analysis, even with terabytes of data, we are still struggling forecasting the unexpected “black swans”. This talk will illustrate several winning approaches in rare events detection, like predicting the number of insurance claims per day.

Key Takeaways

– How to process with unbalanced datasets
– Can we forecast what we have not seen?
– Rare events prediction

————————————————————————————————————————————————————

Speaker Bio

I have a PhD in Physics and Mathematics from Moscow State University, Russia. Working with machine learning in data science field from 2009. Since that delivered AI/ML solutions (including deep learning, big data analysis, etc.) within digitalization, machine learning applications and data driven decision support for renewable energy sector, electricity consumption prediction and forecasting, fishing industry, shipping and logistics, IoT data processing, health sector, and fish farming biology.

From 2015 employed as Head of Data Science at NORCE (former UniResearch), Bergen

And from 2019 serving as Lead Data Scientist, StormGeo, Bergen

In 2018-2019 received support from Norwegian Research Council as inventor and PI (primary investigator) in two innovation projects within digitization and data science application for fish farming and renewable energy.

May 24 @ 13:55
13:55 — 14:25 (30′)

Stage 2 | Analytics Modelling and Innovation

 Program Day 2, Alla Sapronova – Lead Data Scientist | StormGeo