Mind the Gap! Hype-Cycle versus Business Reality in Data Science (a forecasting perspective)

Abstract – The buzz of the “Big Data” revolution had been unnerving CIOs for more than half a decade, when it was suddenly dropped from the Gartner hype-cycle in 2016. Although businesses are still struggling to store more and more data, the emphasis has shifted to making better use of the data through new Data Science algorithms and new Business Applications. Machine Learning, (self-service) Advanced Analytics and Neurobusiness have entered the Gartner hype-cycle, promising the future breakthroughs. However, in Time Series Prediction (an area of Data Science growing in importance with more data gathered continuously over time), aka Forecasting, the corporate reality looks rather different. The elusive crystal ball into the future is often powered by simple and elderly algorithms, many of them around since the 1960s or earlier. Industries as software vendors are slow to adapt machine learning, or indeed even anything contemporary from the 90s. In our presentation, we show evidence from an industry survey of 200+ companies and their reality of algorithms used, and measure the substantial gap between research and practice. To contrast this, we showcase a selection of state-of-the-art algorithms available in Data Science for time series today, from Neural Networks to Support Vector Machines and from Random Forests to Boosting, and how they could be applied to time series Analytics to drive a revolution. We will give examples how these have been implemented by a few industry though-leaders, from Electricity & Utilities companies to Call-Centres, Manufacturers and Container Shipping lines, who are bridging the gap to lead the hype-cycle onwards.

Biography – Sven F. Crone is an Assistant Professor in Management Science at Lancaster University, UK, where his research focuses on business forecasting with artificial intelligence and time series data mining. As the co-director of the Lancaster Research Centre for Forecasting he and his team regularly take state-of-the-art research and apply it in corporate practice for Fortune 500 companies. Recent projects include new product textile forecasting using time series clustering, artificial intelligence for consumer goods forecasting, and data mining approaches in call centre and energy forecasting. Sven regularly provides in-house training courses for companies, and has been a regular speaker at numerous industry conferences, including keynotes at SAS Analytics conferences A2013 USA, A2012 Europe and Predictive Analytics World PAW Europe. He co-chaired the first SAS A2011 in Orlando and PAW2016 and PAW2015 in London, serves on the SAS UK Academic Advisory Board and is chair of the IEEE Computational Intelligence Society Industry Liaison activities. To close the widening gap between research and industry, he recently founded the spin-of iqast which pioneers intelligent algorithms for forecasting.