PERBAIKAN KEAKURATAN KLASIFIKASI POTENSI SATPAM DENGAN MENGGUNAKAN JARINGAN SARAF TIRUAN TERHADAP METODE ANALISIS DISKRIMINAN KLASIK

Abstract

Discriminant analysis is a statistical technique that uses information available in a set of independent variables to classify the value of a descrete or categorical dependent variable. Depending on the research that have been done discriminant analysis can not classify the pattern with perfect accuracy (100%). This study aimed to improve the accuracy of discriminat analysis in classificationing the security potencial with neural network. This study will show the comparation between neural network and discriminant analysis in their work on classifiction the security potencial. The security potencial will be measured with capability, behavior, and skill variable. Three layer neural network was trained with the back propagation algorithm, input layer consist of 3 neurons, hidden layer consist of 60 neurons and output layer consist of 1 neuron, the value of momentum was 0,6 dan the value of initial learning rate was 0,4. After the simulation of neural network was done, the reseacher find that neural network has a better work product than analisys discriminant in classificationing the security potencial.