Abstract:
Geographic Information System has tremendous applications in election management. This study represents and demonstrates the development of a population-weighted & fixed point algorithm for gerrymandering free electoral redistricting and its implementation in ArcGIS using python script. An algorithm based on the mathematical framework has been proposed, ensuring that the new legislative districts so created as a result of delimitation under this algorithm are not only spatially compact and contiguous but demographically homogeneous with nearly equal population size in all districts. Optimization of the districting problem has been achieved by incorporation of optimization constraints in place of the objective function used earlier worldwide. The proposed algorithm generates homogeneous districts by using a single demographic attribute. In addition the algorithm also automates the constituency report against each legislative district created and generate delimitation summary comprising the necessary information about all newly created legislative districts in pdf format. The Python script for implementation of the proposed model has been put in a nutshell in ArcGIS tool named as “The GiraLD”. The GiraLD require with two feature layers as input i.e. the census data layer with population in its attribute table and the respective layer of generator points for desired legislative districts and gives the feature layer of new boundaries formed after dissolving the resultant clusters of basic units into single entity plus the pdf reports as its output. The pilot run on the study area shows that the GiraLD based on the said algorithm can deliver the results with upto 4.4% percent of population tolerance under normal circumstances when shape of the area of interest (AOI) comprising basic population blocks is non-complex and achieves the tolerance of about 7.5% in a worst case scenario when shape of the AOI is too xii complex. The ArcGIS tool developed can be implemented for unbiased redistricting of the whole country.
