Evolution of the SME loan portfolio in Ecuador: a Holt-Winters approach and the Extreme Learning Machine network

Urdaneta Montiel A.J.; Zambrano Morales Á.A.; Condori Cari L.W.; Morales Vergara J.R. · Encuentros (Maracaibo) (2025) · DOI: 10.5281/zenodo.15467671

Resumen

This mixed-method research focused on predicting the evolution of the portfolio of credits to SMEs in Ecuador. The Holt-Winters model and the Extreme Learning Machine network, combining econometric models and neural networks, were employed, along with geospatial analysis at the provincial level. The model exhibited optimal fitting to real data, effectively capturing 93% of their variability, and demonstrated efficient forecasting with slightly superior performance compared to other analyzed models. This outcome holds crucial importance for decision-making and financial resource planning dedicated to Ecuadorian SMEs.