Klasifikasi Usaha Mikro Kecil Menengah Menggunakan Jaringan Syaraf Tiruan Backpropagation


  • T Hardoyo Universitas Sanata Dharma
  • Eko Hari Parmadi Eko Universitas Sanata Dharma




UMKM, backpropagation, accuracy, classification


Nowadays, the development of Usaha Mikro Kecil Menengah (UMKM) is quite rapid. Updating data is very necessary to find out how far the development of UMKM is every year.UMKM are divided into three criteria, namely: micro, small, and medium. The problem is to determine the criteria for an UMKM based on several attributes such as: No, District, Kelurahan, Company Name, Owner's Name, Address, Telephone/HP, Type of Business, Number of Employees, Assets, Turnover, Year of Establishment, and criteria as labels. This takes a long time for the Government to determine the criteria for UMKM. This study uses data from the 2018 UMKM in the city of Bandung. ThisUMKM data will be used for classification so that criteria data can be obtained faster. The classification method used is backpropagation. The data used in this study amounted to 5219 data with 12 attributes and 1 criteria label. The 12 existing attributes are then selected into 4 attributes according to the attribute ranking. Data testing using 3-fold cross validation resulted in an accuracy of 98.4294% with the most optimum network architecture: 30 neurons, using two hidden layers, logsig activation function, trainlm training function, input layer 4 nodes and output layer 2 nodes.