Discussion
Results obtained thanks to the map conflict with results of the scatterplot and this might be explained by the way distances were calculated in the distance matrix. For each point of the grid, the distance to every polluted sites is calculated and the mean value will be given in the distance matrix. This way of calculation biaises the results because a point located really closed to a polluted sites (location A for instance), containing therefore high value of DCGP, might be located very far from the rest of the polluted sites and have a really high mean value in the distance matrix. In contrary to points with low DCGP value which might be more equidistants to polluted sites and have a lower mean value.
One solution in order to confirm the spatial correlation shown on figure 1 would be to calculated the mean DCGP value of the 5 (this number depends on the weighted scheme chosen by the operator) nearest neighbors for each polluted sites centroïd and see if thoses values confirm the original assumption concerning Vernier's commune architecture. Thanks to the ditance matrix, DCGP values of the five nearest neighbors can be determined. As they are 82 polluted sites, a table with all those mean values correponding to each sites would be too massive and not very relevant. However an average value could be calculated in order to give an order of magnitude of the DCGP value of the closest points to polluted sites. A value of 523.53 meters has been calculated here, corresponding to bright orange points on figure 1. It belongs to approximately to the middle of the DCGP value range (11-1105) and doesn't really represent the spatial correlation between the two studied parameters.
Here again, a further analysis is needed concerning the type of pollution sites. More polluted sites (landfill for instance) will tend to be located more in the periphery of the city whereas less polluted ones (exploitation area) will be closer to the city center and its activity. To illustrate this fact, for the landfill at location A (cf Fig 1) and the wastewater treatments plants at location B (cf Fig 1) , the average DCPG values are indeed equal to 951.97m and 858.48 meters respectively .
Conclusion
Regarding to the results, it might be diffult to assess if the proximity to polluted sites and the distance to general practitioner are spatially correlatted or not. This confusion is mainly explained by the great diversity of polluted sites and their various seriousness of pollution. However, the common architecture of a city tends to be apllied in the Vernier case where highly polluted sites are located far from city's activities and less polluted ones are located closer but still in the periphry of the city center.