The usefulness and the efficiency of the kth nearest neighbor classification procedure are well known. A less sophisticated method consists in using only the first nearby prototype. This means k=1 and it is the method applied in this paper. One way to get a proper result is to use weighted dissimilarities implemented with a distance function of the prototype space. To improve the classification accuracy and to minimize the number of parameters, functions that shape the relation of the dissimilarity with the prototypes, the features or the class belonging are proposed. Benchmark tests are also presented and the worth results encourage us to continue developing this new proposed weighted model.