International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Publications Copernicus
Download
Citation
Volume XLII-4/W19
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-4/W19, 455–462, 2019
https://doi.org/10.5194/isprs-archives-XLII-4-W19-455-2019
© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-4/W19, 455–462, 2019
https://doi.org/10.5194/isprs-archives-XLII-4-W19-455-2019
© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

  23 Dec 2019

23 Dec 2019

GEOSPATIAL AND CLUSTERING ANALYSIS OF DENGUE CASES USING SELF-ORGANIZING MAPS: CASE OF QUEZON CITY, 2010–2015

J. J. Valles1, C. Perez1, and A. C. Blanco1,2 J. J. Valles et al.
  • 1Dept. of Geodetic Engineering, University of the Philippines – Diliman, Quezon City, Philippines
  • 2Training Center for Applied Geodesy and Photogrammetry, University of the Philippines – Diliman, Quezon City, Philippines

Keywords: Dengue, Self-Organizing Map, Clustering, OLS Regression

Abstract. Dengue is the most rapidly spreading disease in the world with more than 30% of the world’s population at risk of contracting dengue. In 2016, more than 375,000 suspected cases of dengue were reported from the Western Pacific Region, and more than half of these were reported by the Philippines. Dengue virus inflicts significant health and economic burden to the Philippines. Thus, it is important to improve the country’s current schemes for dengue surveillance and response thru better understanding and knowledge on the development of dengue. In this research, geospatial and clustering analyses of dengue cases in Quezon City through GIS and self-organizing maps (SOM) were performed. Two clusters were generated for each clustering method. After clustering the barangays, the coefficient of determination increased for most scenarios compared to the OLS regression of the ungrouped data. The R2 values for the regression of whole Quezon City dataset ranged from 0.364 to 0.671, while it ranged from 0.468 to 0.839 for the SOM-clustered dataset. On the other hand, for the k-means-clustered dataset, R2 values ranged from 0.395 to 0.945. Moreover, GWR models’ adjusted R2 values ranged from 0.675 to 0.876. Common predictors among the different regression models are the informal settlements and very low residential areas. Based on the significant predictors identified and the trend of the dengue cases, SOM produced more logical classification than the GIS Grouping Analysis. Although SOM takes a longer time compared to the GIS Grouping Analysis, SOM is easier and simpler to implement.