A Comparative Analysis of LightGBM and ES-RNN Applied to Time Series Forecasting

Simple item page
AudienciaPúblico en generales_ES
CoberturaMéxicoes_ES
Fecha de ingreso2026-10-05T16:33:53Z
Fecha de publicación2026-01-01
ResumenAdvancements in machine learning are reshaping time series forecasting, with the M competitions serving as crucial benchmarks. The M4 and M5 competitions highlighted the strengths of methods like ES-RNN and LightGBM, respectively. This study compares the M4 competition winner (ES-RNN) and a LightGBM forecasting approach inspired by its prevalence among top M5 solutions. We evaluate their performance on a challenging subset of 2724 daily frequency time series from the M4 dataset. Using multiple error metrics (sMAPE, MAE, MAPE, RMSE), our results indicate that LightGBM outperforms ES-RNN in a slight majority (approx. 54–55%) of the series. However, a key finding is that LightGBM’s superior performance is conditional and strongly linked to a discernible trend in the time series. LightGBM demonstrates a significant advantage and higher forecast correlation for series with clear upward or downward trends. Conversely, the hybrid ES-RNN model exhibits greater robustness for series lacking strong trends or displaying higher variability. We introduce a quantitative trend measure and employ non-parametric statistical tests to statistically validate the significant impact of the trend on the correlation between LightGBM forecasts and actual values. We conclude that while LightGBM is highly effective, particularly for trended data common in domains like retail sales, its broader application requires careful consideration of the underlying data characteristics. Understanding the time series trend is crucial for practitioners selecting between these powerful forecasting methods.es_ES
Doihttps://doi.org/10.1007/978-3-032-13497-4_1es_ES
URIhttps://riuat.uat.edu.mx/handle/123456789/5643
Idiomaenes_ES
EditorialSpringer Science and Business Media Deutschland GmbHes_ES
RelaciónIntelligent Systems Reference Libraryes_ES
URL relacionadohttps://doi.org/10.1007/978-3-032-13497-4_1es_ES
DerechosAcceso restringido / Suscripción (Metadatos de producción científica)es_ES
Licenciahttp://purl.org/coar/access_right/c_16eces_ES
FuenteIntelligent Systems Reference Library
Palabra claveForecastinges_ES
Palabra claveLightGBMes_ES
Palabra claveMachine learninges_ES
Palabra claveTime serieses_ES
Palabra claveTrendes_ES
TítuloA Comparative Analysis of LightGBM and ES-RNN Applied to Time Series Forecastinges_ES
TipoCapítulo Libroes_ES
ArbitradoHa sido Arbitradoes_ES
AutorMata-Alvarado, Antonio
AutorPonce-Flores, Mirna P.
AutorIbarra-Martínez, Salvador
AutorTerán-Villanueva, Jesús David
AutorLaria-Menchaca, Julio
AutorMata-Alvarado, Antonioes_ES
AutorPonce-Flores, Mirna P.es_ES
AutorIbarra-Martínez, Salvadores_ES
AutorTerán-Villanueva, Jesús Davides_ES
AutorLaria-Menchaca, Julioes_ES
InstituciónUniversidad Autónoma de Tamaulipas
InstituciónUniversidad Autónoma de Tamaulipases_ES
Rango de páginas3-40es_ES
URL relacionadahttps://doi.org/10.1007/978-3-032-13497-4_1
Tipo de artículoIndexado
Tipo de artículoIndexadoes_ES
Volumen283es_ES

Files

Collections

Documentación Cientifica