<![CDATA[Drug abuse is one of the major problems for Indonesian country. This is due to drug users can cause psychiatric disorders, health and even death. From various surveys conducted, the number of cases of drug users increase every time. In this study, the research will try to do initial identification for users with Learning Vector Quantizationmethod. This study uses the data of drug users from Badan Narkotika Nasional (BNN) in Malang Regency. From 119 data, there will be divided into 3 parts. There are 4 data for initial weight vectors, 103 for training data and 12 for test data. Then, in this data have 16 parameters and 4 classes. In this study, 6 tests were performed, resulting in 0.1 for the learning level with an average score of 74.8%. Then, 0.9 for a learning rate multiplier with an average rating of 79.8%. Then, 0.01 for the minimum level of learning with an average grade. The amount of training data is 60% with average value. The maximum of 14 lawer iterations with average values. Then, for LVQ training stops on which best condition is at maximum iteration 14. Therefore, the minimum of learning rate condition can be ignored. From those results, the average of final accuracy after testing with K-Fold Cross Validation is 78.4%.]]>
Copyrights © 2018
