Event Details

When modelling time-series data there is often a tendency to focus on developing the ‘best’ model for a given situation without considering the inherent fragility of the model itself.

In practice, significant forecasting errors typically come from structural changes or changes in the availability of data at the point of forecast. What if a competitors store opens up next door, or your model relied on a data point from t-x but t-x data does not arrive in time? Extended periods of sub-optimal or incomplete forecasts can follow before the next model is built, often with considerable investment in time and energy but the revised models are no less fragile.

Model accuracy is important, so for a forecasting system to be resilient and manageable at scale the models must capture new information from the data as soon as possible, whilst also providing the transparency and granularity required to enable users to understand the specific impact of these changes in real time.

This webinar will explain why Tangent Works focussed on automating the creation of time-series machine learning models and the implications that their InstantML technology has in enabling resilient forecasting philosophies.

Mike is a Director of Tangent Works UK. The Tangent Information Modeller by Tangent Works builds machine learning models with a single pass of the data in just a few moments. Since studying Physics at The University of Manchester, Mike has worked with various global technology companies before returning to The Alliance Manchester Business School to complete his MBA.

mike.thomas@tangent.works

www.tangent.works