<![CDATA[One of the major problems that Indonesia has is unemployment. It happens because of the low number of the job vacancy meets the high number of the graduate and job seeker. The unbalanced number of the job vacancy to the graduate and job seeker causes an increasing in the unemployment rate in Indonesia. The high rate of unemployment will inevitably give a direct or indirect impact to the poverty, crime, and other social issues. However, according to the survey conducted by BPS, Indonesia's unemployment rate was even increasing from the year of 2014 to 2015. It would really help the Indonesian government to make planning, program, or policy related to the job vacancy by being able to predict the number of unemployment. In addition, the data result from the prediction could be used as the measurement of the success of government's previous programs. Backpropagation is one of the methods that used to predict. This research works on the optimation of weights initialization in Backpropagation using Bayesian method that is a modification of the Kalman filter. According to the test result done in this research, the lowest value of Average Forecasting Error Rate (AFER) is 2,1003%. From that result, the combination method between Backpropagation and Bayesian has better accuracy rate than Backpropagation method with random initialization that has 2,5793% lowest value of AFER, but need a lot of iterasion. The sistem of this research can predict some future of years immediately but the most optimal result is the first year while in subsequent years the result of the prediction is more inaccurate.]]>
Copyrights © 2017
