MetaARIMA: Automatic Configuration of ARIMA using Classifier Chains

Vitor Cerqueira, Ricardo Inácio, Carlos Soares

Submission Type: Short paper

Tldr: Using meta-learning to model features extracted from time series and estimations of performance from diverse ARIMA configuration to find the optimal one, in an efficient and effective manner by using a classifier chain, top-k selection, successive halving and AICc-based final selection, outperforming AutoARIMA.

Submission Number: 1