Publicación: Forecasting Airborne Pollen Concentrations through Random Forests
Cargando...
Fecha
2016-09-01
Autores
Editor/a
Director/a
Tutor/a
Coordinador/a
Prologuista
Revisor/a
Ilustrador/a
Derechos de acceso
info:eu-repo/semantics/openAccess
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Nacional de Educación a Distancia (España). Escuela Técnica Superior de Ingeniería Informática. Departamento de Inteligencia Artificial.
Resumen
Poaceae is the largest family of monocotyledonous flowering plants, known as grasses and considered to be one of the most important aeroallergens in Europe. The increase of allergy cases and the severity of the reactions motivates the prediction of atmospheric concentrations in order to minimize the exposure to risky pollen levels. Also, it is of large interest for clinical institutions in order to apply preventive measures and plan in advance the implications of an increase number of allergy patients. Phenological and meteorological parameters characterize the stages of vegetation development during the growing season. Thus, they can be potentially related to the biological definition of plant phenology. In this thesis, time series of airborne pollen concentrations and meteorological variables measured at the region of Madrid were used to predict risk pollen concentrations for patients. Detailed relationships were established between future airborne concentrations, meteorological data, and flowering states derived from the inner information of the underlying data via computational intelligence models. Therefore, these data were used to develop predictive models for a range of forecast horizons. The proposal will be beneficial to the medical fields related with allergies affections planning and treatment, demonstrating that computational intelligence holds a great potential for aerobiology. In this research, we demonstrated several novel approaches that significantly contribute to the field of aerobiology including: (i) developing a computational intelligencebased model to predict risk concentration levels, and consequently the start and end of pollen season, for long term horizons up to 6 months, on which none of previous works succeed to obtain satisfactory results, (ii) identifying and characterizing the most influential factors which induce the presence of high airborne concentrations for a given set of horizons, using an assumption-free approach which supports biometeorological findings from other authors. The findings of the research are related to producing more accurate prediction models and providing a comprehensive analysis of the relationship of airborne concentrations with various meteorological parameters. As a product of this study, a version of Chapter 3 of this thesis was accepted and presented at the International Work Conference on Time Series Analysis (2016)1, and it has been selected to be extended and submitted as a book chapter in the Springer series Contributions to Statistics2. A version of Chapter 5 has been published as a full paper in the International Journal of Biometeorology3 and Chapter 4 is in preparation to be sent to an international journal for publication.
Descripción
Categorías UNESCO
Palabras clave
Citación
Centro
Facultades y escuelas::E.T.S. de Ingeniería Informática
Departamento
Inteligencia Artificial