Perdomo García, Cristian Raúl

Profile Picture

Filters

20231
Reset filters

Settings

Citations

Publication Search Results

Now showing 1 - 1 of 1
  • Thumbnail Image
    Publication
    Monitoreo de calidad de agua de ríos en Puerto Rico usando modelos aditivos generalizados mixtos (GAMM)
    (2023-06-26) Perdomo García, Cristian Raúl; Macchiavelli, Raúl E.; College of Arts and Sciences - Sciences; Santana Morant, Dámaris; Lorenzo González, Edgardo; Department of Mathematics; Sotomayor Ramírez, David R.
    Mixed linear models are a generalization of linear models and are widely used in different branches of science, as they allow for statistically modeling possible correlations in the data. However, in many cases, the data is not linearly related and although classical methodologies can be used, there are alternatives such as generalized additive mixed models. In these models, the smooth function that relates the mean value of the response variables y to the predictor variables is constructed using smooth function bases. This work studies the properties of this type of model and applies it to the analysis of data from 55 rivers in Puerto Rico with the goal of estimating trends over time in water quality parameters such as dissolved oxygen and total nitrogen concentration. The data was obtained from the records of the United States Geological Survey (USGS), compiled and consolidated by Dr. Gustavo A. Martinez and Miguel Vázquez of the Agricultural Experiment Station at the University of Puerto Rico. The available data covers the period from 1969 to 2021. The mixed additive models were fitted using the gam function of the mgcv library in (R v. 4.2.2). The longitudinal nature of the data was incorporated into appropriate random effects, and classification variables were included to study differences in river conditions. Smooth functions were fitted using P-splines to study these effects over time. The studied interactions explained the observed trends and yielded specific predictions of the condition, which were compared under the construction of functions that allowed generating appropriate contrast matrices. This approach also enabled the assessment of statistically significant changes in trends due to specific breaks in time and predictions of current trends in water quality parameters.