To populate Scholarly, sign in here .


Title Autologistic Models with Autocorrelated Errors for an Aquatic Phenomenon
Posted by Bernadette Tubo
Authors Bernadette S. Fermin and Erniel B. Barrios
Publication date 2002
Journal The Philippine Statistician
Volume 51
Issue 1-2
Pages 41-50
Publisher Philippine Statistical Association INC
Abstract Spatial and autocorrelation structures are hypothesized to be important components in addition to the hydrological variables to describe the stage of an aquatic phenomenon indexed by a binary response variable. Two models that incorporate spatial and autocorrelation parameters to an ordinary logistic model (autologistic) are proposed. An algorithm similar to the backfitting routine is proposed. The ordinary logistic model is used as a benchmark for comparison. Results indicate that the autologistic model is superior to the other two models in terms of its ability to recognize the stage of a binary event. Inclusion of the spatial parameter in autologistic model is a significant improvement in the prediction of binary events.
Index terms / Keywords autocorrelation parameter, backfitting algorithm, binary response variable, harmful algal bloom (HAB), maximum pseudolikelihood, sampling stations, spatial parameter