Forecasting challenges and MIY: Model It Yourself - with ML Legobricks
Details
💥Welcome to our third meetup for 2021💥
In this meetup we will start one hour earlier than usual, at 18:00 and we will have two talks.
Spyros Makridakis will talk to us about the challenges in the field of forecasting and how the M competitions guided the theory and practice of forecasting
Sai Ma will talk to us about ML-legobricks, a machine learning project used internally at Klarna that performs MLOps tasks by giving you the ingredients and allowing you to cook your model as you want 😃
Grab your beer 🍺 or tea 🍵 at home and join us virtually through the zoom link. We look forward to seeing you there.
Agenda
18:00 - 18:40 Talk #1 Spyros Makridakis: Forecasting challenges and the contributions of the M competitions
18:40 - 18:50 Q&A
18:50 - 19:30 Talk #2 Sai Ma: MIY: Model it Yourself with ML Legobricks
19:30 - 19:40 Q&A
19:40 - 20:00 Networking
Talk #1 - Forecasting challenges and the contributions of the M competitions
Abstract
Predictions are needed for planning, scheduling, supply chain and similar decisions concerned with future, uncertain events and is the role of forecasting to provide such predictions to allow making these decisions as rationally and efficiently as possible. This talk discusses the many challenges facing the field and describes how the M competitions have offered objective, empirical evidence to guide the theory and practice of forecasting towards methods that are both more accurate while, at the same time, estimating uncertainty more realistically.
About the speaker
Spyros Makridakis is a Professor at the University of Nicosia, where is also a director of its Institute For the Future (IFF) and the founder of the Makridakis Open Forecasting Center (MOFC). He is also an Emeritus Professor at INSEAD. He has authored, or co-authored, twenty-four books and more than 270 articles. His book Forecasting Methods for Management, 5th ed. (Wiley) has been translated in twelve languages and sold more than 120,000 copies while his book Forecasting: Methods and Applications, 3rd ed. (Wiley) has received close to 6,000 citations. Professor Makridakis was the founding editor-in-chief of the Journal of Forecasting and the International Journal of Forecasting and is the organizer of the renown M (Makridakis) Competitions that for the last 40 years have fundamentally influenced the theory and practice of forecasting. His papers/citations can be found in his Google scholar profile: scholar.google.com/citations?user=hPpgXPMAAAAJ&hl=en.
Talk #2 - MIY: Model It Yourself - with ML Legobricks
Abstract
ML-legobricks is a machine learning project that includes one code-sharing library, one agreement with Sagemaker and one cookiecutter template. It aims to unify the model building cycle, creating reproducible pipeline steps using Docker, and essentially build a model by one click away, not only providing you ingredients and recipes but also allowing you to cook your model as you want. We have this project onboarding internally in the Klarna fraud prevention team, and plan to push it to be adopted globally within Klarna.
About the speaker
Sai Ma is a senior data scientist in Klarna, specifically responsible for fraud detection in the Fraud Prevention Team. Before Klarna, she had working experience in content analysis, recommendation systems and CLV in the largest sportswear manufacturer in Europe. Sai Ma has studied mathematics (Geometric Analysis) at Polytechnique, Paris, France, and has been an international butterfly living/studying/working in 5 countries. She believes that to become an outstanding data scientist, there are 3 skill baskets to fill: deep understanding in theory, practical engineering skills and business storytelling. She is very excited to be here and is looking forward to sharing ML-legobricks in PyData Cyprus.
Code of conduct
Forecasting challenges and MIY: Model It Yourself - with ML Legobricks