Online Time Series Prediction with Missing Data

We consider the problem of time series prediction in the presence of missing data. We cast the problem as an online learning problem in which the goal of the learner is to minimize prediction error. W