Business Forecasting with R. Are we there yet?

7 May 2019 15:25

R is becoming increasingly popular in companies, offering a wide variety of state of the art solutions for various analytics problems. When it comes to forecasting algorithms and solutions, in contrast to many examples of commercial software, open source software, such as R, implement the latest developments. This workshop looked into the opportunities that R offers to companies.

The first talk by Nikos Kourentzes on “Open source, R and business practice: the state of play”. The talk illustrated the rise of popularity of open source software over the recent years but also noted that academia is still its biggest user. While open source software often has more cutting-edge tools on offer, it falls short of offering complete solutions that can be readily used by businesses, such a integrated forecasting support systems. The importance of interface for the users, as well as the ecosystem to inegrate data sources, outputs and reports should not be understated. This leads for opportunities for co-implementation of proprietary software and open source forecasting. The slides are available here.

In the second, Ivan Svetunkov highlighted solutions to solve business forecasting problems. The Centre for Marketing Analytics and Forecasting is at the forefront of research in business forecasting and its members maintain a significant number of R packages (full list available here). This packages are part of the mission the Centre has, to make research easy accessible to practice. In the talk Ivan demonstrated the usage of those packages within a simple tutorial. The slides of the talk are available here and R scripts can be found here.

After the coffee break, three talks from practitioners illustrated the use and implementation of R code within business environments. First, Dmitrii Ishutin provided an insights on how EDF energy forecasts: What R packages they use in daily operation and how RMarkdown helps them to create interactive reports of their forecasts. The slides are available here.

Andrea Liotta then provided insights on how GfK moved from a human to a data-driven approach with use of R. One of the challenges for them was to get a fully automated forecast pipeline. Andrea shared ways that helped GfK to standardise code and assure quality. Those steps are necessary as the flexibility of R to facilitate various coding practices can at times hinder reliable deployment and operations for businesses. Download slides.

Finally, lektra Skepetari highlighted how R facilitates predictive analytics tasks at Heathrow. Her team particularly values the availability of state-of-the-art methods and the visualisations capabilities. Again, RMarkdown is used for reporting. However, Ilektra also pointed out challenges of R with regards its memory management computational speed limitations. The slides are available here.

Back to News