Posts
6 Common Pitfalls for Forecast Evaluation
A topic I covered last year in some talks and papers are the “6 common pitfalls for forecast evaluation”. I’m discussing what are the most typical mistakes people new to forecasting would make. So this is relevant, for example, for Data Scientists that may not have any specialised training in forecasting, but in ML and Stats. It is a more lightweight take on the same topic covered in our quite detailed and more formal full paper here.María Zambrano (Senior) Fellowship at University of Granada, Spain
I’m very fortunate as I have been offered a María Zambrano (Senior) Fellowship at the University of Granada, Spain. My alma mater where I did the PhD. I’ve taken on this role now, and for the next 2.5 years I will be on a research position with minimal teaching to be able to focus further on my forecasting research. I’ll stay in connection and continue collaborating with my colleagues and friends at Monash University, where I now hold the appointment of an Adjunct Senior Research Fellow.Visiting Researcher at Meta
Since about a week I’m on a sabbatical for 6 months in the Infrastructure Data Science and Engineering team at Meta Platforms Inc. in Menlo Park, California. I’m fortunate that Zeynep Erkin Baz, Dario Benavides, and Ashish Kelkar have given me this opportunity to work with them in their great team. I’m looking forward to tackling interesting forecasting problems and learning a lot more about forecasting and other things along the way.FUZZ-IEEE Explainable Energy Prediction Competition
My PhD student Dilini Sewwandi Rajapaksha and I made first place in the FUZZ-IEEE Explainable Energy Prediction Competition. The winners were announced at the IEEE International Conference on Fuzzy Systems, Luxembourg, 2021.
We proposed a novel algorithm to provide Local Interpretable Model-agnostic Rule-based Explanations for Forecasting, a paper and more descriptions will be available hopefully soon.
IEEE-CIS Technical Challenge on Predict+Optimize for Renewable Energy Scheduling
We’ve just launched the IEEE-CIS Technical Challenge on Predict+Optimize for Renewable Energy Scheduling, with US$20k in prize money. You’ll need to forecast energy demand and solar power production for a couple of buildings on the Monash Clayton campus, and then use the forecasts to optimally schedule lectures and battery charging/discharging. We’re looking forward to your great submissions!
More information is here
Special session at Virtual ISF2021 on Recent Advances in Global Modelling for Forecasting
We’ve been organising an invited session at Virtual ISF2021 on Recent Advances in Global Modelling for Forecasting. Details on part 1 and 2 are here and here Session 1: Time series feature embedding for forecasting with deep learning Speaker: James Nguyen A Look at the Evaluation Setup of the M5 Forecasting Competition Speaker: Hansika Hewamalage Dependency Learning Graph Neural Networks for Multivariate Forecasting Speaker: Abishek SriramuluOur new time series repository: Forecastingdata.org
We have put together a new time series data repository for forecasting. It is dedicated to sets of series for cross-learning/global modelling, and to single but very long series (the longest ones over 7 million data points). We were also not too happy with any existing data format, so we developed a new one and we think it is quite useful and versatile.
Paper link: https://arxiv.org/abs/2105.06643
Website: https://forecastingdata.org
IEEE-CIS Competition
Congratulations to my PhD students Kasun, Hansika, and Rakshitha for their 4th place in the IEEE-CIS Technical Challenge on Energy Prediction from Smart Meter Data
Well done! Kasun describes some of the methodology here
Also congratulations to Alex Dokumentov for his 6th place (3rd in terms of accuracy).
Neural Prophet released
We have released the Neural Prophet software, a reimplementation of “prophet” in PyTorch. It is joint work with Facebook and Stanford University.
Announcements from Facebook are here and here
The code is here
Tutorial at ACML2020 on Forecasting for Data Scientists
I’m giving a 2.5 hour tutorial at ACML2020 on Forecasting for Data Scientists, covering all about forecasting that Data Scientists and Machine Learners should know.
Get additional information here