Identificación y simulación de patrones de siembra de cultivos ilícitos de coca mediante minería de datos y autómatas celulares

Abstract

Alternatively in the identification of coca cultivation in Colombia, it is proposed in this research a methodology based on remote sensing techniques, data mining, fuzzy logic and cellular automata to model the phenomenon of coca planting in the municipality of Miraflores in Guaviare. The proposed methodology can identify the underlying rules that determine the probability of existence of illicit coca crops using the strengths of the techniques of spatial data mining and fuzzy logic. These rules are the input to program the cellular automata so that the phenomenon can be modeled based on the fact that each pixel in the image is a cell controller. The methodology includes four phases i) Analysis and processing of information collected ii) Identification of variables used iii) construction of fuzzy logic model and finally, iv) construction of cellular automata model capable of simulating the planting of coca cultivation Final results after validation give a reliability above 75%, measured by comparing pixel to pixel "image result" with the conventional interpretation of coca crops in the study area