Automated Dynamic Machine Learning for time based forecasts


Research in efficient data exploitation continues to demonstrate major breakthroughs world-wide, especially with the advent of large volumes of available data. Unfortunately, most of this research is still restricted to researchers and therefore medium and small company are unable to take advantage of these advanced technologies due to a combination of lack of internal skills and the large upfront costs and time necessary to master these new data technologies.

We aim to design innovative services and products to help companies by facilitating their access to efficient data exploitation technologies and therefore to help them stay ahead of their competitors by endorsing the right, cost-effective, data-driven approach. The data analysis domain addressed by Predictive Layer is Time Series analytics. In this respect, impressive results have already been achieved in the case of energy consumption prediction. 

We shall base this approach on the automated analysis of the topology of the problem that is being treated. In other words, it will create, dynamically during the learning process, the best features of representation of the time series, based on previous explored and tested features.

Collaboration: Predictive Layer SA