Imputation techniques applied in a maximum monthly precipitation data in the central zone of Boyacá

Precipitation directly affects the water supply of river basins and its prediction becomes the main objective in different investigations. However, historical records often show missing data due to instrumental, technical or human drawbacks. This limitation must be solved to avoid errors in subsequent Analysis. This proposal deal with a similar problem for a data set about precipitation collected in the central part of Boyacá along the years 1974-2013. The performance of the imputation mechanisms of loss MCAR, MAR and MNAR was evaluated. All of them were implemented each one under either a multiple imputation with a random approach based on an allocation by the K-Nearest Neighbors method with spatial focus and an imputation by the Kalman smoothing method time focused approach. We measured the convergence of the descriptive statistics of the imputed value and the original value, and additionally, we compared the graphical adjustments and their probability distributions. Amelia was suggested as a better performance of imputation technique jointly with a gamma distribution associated to the missing data.

• Tipo de recurso: http://purl.org/coar/resource_type/c_6552
• Fecha de publicación: 2020
• Institución: Universidad Pedagógica y Tecnológica de Colombia
• Repositorio: RiUPTC: Repositorio Institucional UPTC
• Idioma: spa
• OAI Identifier: oai:repositorio.uptc.edu.co:001/12311
• Acceso en línea: https://revistas.uptc.edu.co/index.php/ingenieria_sogamoso/article/view/12209; https://repositorio.uptc.edu.co/handle/001/12311
• Palabra clave: Multiple imputation; precipitation; R-software; temporal series; Boyacá; Imputación múltiple; Precipitación; R; series temporales; Boyacá
• License: http://purl.org/coar/access_right/c_abf53