[2602.03912] Echo State Networks for Time Series Forecasting: Hyperparameter Sweep and Benchmarking

Computer Science > Machine Learning arXiv:2602.03912 (cs) [Submitted on 3 Feb 2026 (v1), last revised 30 Mar 2026 (this version, v2)] Title:Echo State Networks for Time Series Forecasting: Hyperparameter Sweep and Benchmarking Authors:Alexander Häußer View a PDF of the paper titled Echo State Networks for Time Series Forecasting: Hyperparameter Sweep and Benchmarking, by Alexander H\"au{\ss}er View PDF Abstract:This paper investigates the forecasting performance of Echo State Networks (ESNs) for univariate time series forecasting using a subset of the M4 Forecasting Competition dataset. Focusing on monthly and quarterly time series with at most 20 years of historical data, we evaluate whether a fully automatic, purely feedback-driven ESN can serve as a competitive alternative to widely used statistical forecasting methods. The study adopts a rigorous two-stage evaluation approach: a Parameter dataset is used to conduct an extensive hyperparameter sweep covering leakage rate, spectral radius, reservoir size, and information criteria for regularization, resulting in over four million ESN model fits; a disjoint Forecast dataset is then used for out-of-sample accuracy assessment. Forecast accuracy is measured using MASE and sMAPE and benchmarked against simple benchmarks like drift and seasonal naive and statistical models like ARIMA, ETS, and TBATS. The hyperparameter analysis reveals consistent and interpretable patterns, with monthly series favoring moderately persistent re...