Spatial Relationship of Drug Smuggling in Northern Thailand Using GIS-based Knowledge Discovery

Abstract

The number of drugusers has been growing, likely caused by oppressive social conditions. The drug situation in Thailand has changed so that it is no longer a production source. However, Thailand is one of the transit sites for narcotics smuggling. Drug smuggling occurs most recurrently along the border of Northern Thailand by topographic roads. Chiang Mai and Chiang Rai Provinces have been shown to have the highest statistics in terms of drug trafficking. In this investigation, eight districts adjacent to neighboring countries were chosen as the areas of study. These areas are Mae Chan, Mae Fa Luang, andMae Sai located in Chiang Rai Province, as well as Fang, Chiang Dao, Mae Ai, Chai Prakan, and Wiang Haeng situated in Chiang Mai Province. This research studied the spatial relationship of factors related to narcotic smuggling using a data mining-based decision tree technique. The geographic locations of drug trafficking arrests were transferred into a data-mining process in order to assess the spatial relationships among types of exhibited drugs, season, land use, distance from checkpoint and smuggling routes. Drug smuggling risk areas were further predicted using decision tree modeling. The results revealed that the geographic locations of drug trafficking arrests in Mae Chan, Mae Sai, Mae Ai, and Fang Districts were related to the season factor. The distance from checkpoint showed a spatial relationship with drug smuggling arrests in the Chai Prakan District. Narcotic trafficking arrests in Mae Fa Luang were mostly related to land use and type of drug exhibited. Geo-locations of drug smuggling illustrated an independent relationship with smuggling routes. The results retrieved from the prediction-based decision tree method indicated that Chai Prakan, Mae Chan, Mae Sai, Mae FaLuang, Mae Ai, Fang, Wiang Haeng and Chiang Dao Districts were high-risk drug smuggling areas. The precision value of prediction was found to be 0.652. These results could support spatial decision making for national drug smuggling monitoring and surveillance.

Keywords

• drug smuggling
• gis knowledge discovery
• spatial relationship
• decision tree
• prediction