Evolvable fuzzy systems from data streams with missing values: with application to temporal pattern recognition and cryptocurrency prediction

Abstract

Data streams with missing values are common in real-world applications. This paper presents an evolving granular fuzzy-rule-based model for temporal pattern recognition and time series prediction in online nonstationary context, where values may be missing. The model has a modified rule structure that includes reduced-term consequent polynomials, and is supplied by an incremental learning algorithm that simultaneously impute missing data and update model parameters and structure. The evolving Fuzzy Granular Predictor (eFGP) handles single and multiple Missing At Random (MAR) and Missing Completely At Random (MCAR) values in nonstationary data streams. Experiments on cryptocurrency prediction show the usefulness, accuracy, processing speed, and eFGP robustness to missing values. Results were compared to those provided by fuzzy and neuro-fuzzy evolving modeling methods.