Forecasting Under Concept Drift

Forecasting Under Concept Drift

NYU WIRELESS P.I.s

Research Overview

Dennis Shasha’s lab does research in forecasting in environments where the rules can change. Technically speaking, we do research in forecasting for streaming data where there is concept drift. Our work has three aspects:

  1. Algorithms for refusing to make predictions when the underlying predictor is unreliable
  2. Algorithms for efficient machine learning on streaming data with concept drift
  3. Software packages for single variate and multi-variate forecasting