AutoAI-TS: AutoAI for Time Series Forecasting

Large number of time series forecasting models including tradi- tional statistical models, machine learning models and more re- cently deep learning have been proposed in the literature. However, choosing the right model along with good parameter values that performs well on a given data is still challenging. Automatically providing good set of models to users for a given dataset saves both time and efforts from using trial-and-error with wide variety of available models along with all possible parameter optimization. We present AutoAI for Time Series Forecasting (AutoAI-TS) that pro- vides users with a zero-conf system to efficiently train, optimize and choose best forecasting model among various classes of models for the given dataset. With the flexible zero-conf design of AutoAI-TS automatically performs all the data preparation, model creation, parameter optimization, training and model selection for the users and provides them with trained model that is ready to use for their use case. For given data, the AutoAI-TS utilizes wide inventory of models including classical statistical models, Machine Learn- ing (ML) models, statistical-ML hybrid models and deep learning models along with various transformations to create forecasting pipelines. It then evaluates and ranks pipelines using proposed T-Daub mechanism to choose best pipeline. The paper describe in detail all the technical aspects of AutoAI-TS along with exten- sive benchmarking on variety of real world data sets for various use-cases. Benchmark results show that AutoAI-TS with no manual configuration from user, automatically trains and selects pipelines that on average outperform existing state-of-the-art time series forecasting toolkits.