How to Leverage Data for Time Series Forecasting with Artificial Intelligence Models: Illustrations and Guidelines for Cross-Learning

This chapter shows how to use large quantities of data for improving forecasting accuracy of Artificial Intelligence models. Time series forecasting usually proceeds by fitting a model to a given time series. Because time series usually have just a few observations, and AI models require considerable amount of data to train, forecasting by fitting an AI model to each time series in isolation tends to underperform when compared to forecasting with simple models. In this chapter, we analyze a methodology that overcomes the data limitations of AI models in forecasting, therefore enabling the application of the full range of AI techniques to forecasting, which will lead to improvementsImprovement in predictive accuracyPredictive accuracy. The methodology, called cross-learningCross-learning, proceeds by training a single AI model across multiple time seriesTime Series, as opposed to local learning, that trains a model on each time seriesTime Series. We show how cross-learningCross-learning is universally applicable to arbitrary datasets as long as complex models are used, despite its seemingly strong restrictions. We analyze the statistical aspects of cross-learningCross-learning compared to local learning, illustrating how cross-learningCross-learning benefitsBenefit from forecasting many time seriesTime Series and providing guidance on how models and datasets can be tuned in order to maximize accuracy. We give explicit illustrations of how notable time seriesTime Series processes such as logistic growthsLogistic growth, polynomialPolynomials trends or periodic patternsTrend have an equivalent, perfect representation as complex autoregressive processes that are learned automatically by training them on whole datasets. We show an additional benefitBenefit of cross-learningCross-learning in how it can transfer informationTransfer information across series in the dataset, when some time seriesTime Series can be considered to have information that could be helpful to forecast other time seriesTime Series in the dataset. A real case of COVID-19COVID predictions is analyzed to exemplify this transfer of information mechanism. The guidelines we provide are model-agnostic, they either motivate or are directly applicable to the range of AI models currently available.