Adaptive Interval Fuzzy Modeling from Stream Data and Application in Cryptocurrencies Forecasting

Abstract

AbstractThis paper develops an adaptive interval fuzzy modeling method using participatory learning and interval-valued stream data. The model is a collection of fuzzy functional rules whose structure and parameters evolve simultaneously as data are input. The evolving nature of the method allows continuous model update using stream interval data. The method employs participatory learning to cluster interval input data, assigns to each cluster a fuzzy rule, uses the weighted recursive least squares to update the parameters of the rule consequent intervals, and returns an interval-valued output. The efficacy of the method is evaluated in modeling and forecasting daily low and high prices of the two most traded cryptocurrencies, BitCoin and Ethereum. The forecast performance of the adaptive interval fuzzy modeling method is evaluated against classic autorregressive moving average, exponential smoothing state model, and the naïve random walk. Results indicate that, similarly to with exchange rates, no model outperforms random walk in predicting prices in digital coin markets. However, when a measure of directional accuracy is accounted for, adaptive interval fuzzy modeling outperforms the remaining alternatives.KeywordsAdaptive machine learningFuzzy modelingInterval-valued stream dataForecastingCryptocurrencies