Abstract
Beef price forecasting is one of methods to overcome an unstable market price of beef. A forecast method that can be developed is Backpropagation. However, the method requires a high number of data to obtain optimal performance, which is contrary to the availability of observed data. Therefore, we proposed an interpolation technique to overcome incompatibility of data availability, i.e. granularity level of supporting data is in general, whereas granularity level of forecasting need is in detail. The experiments result that the best architecture of Backpropagation is 0.1 as momentum value, 0.01 as learning rate value, and 6 as hidden neuron size that produce 12% of MAPE's value or 88% of testing accuracy. At the same time, the ratio of beef price gap between the actual value and predicted value against the actual beef price is about 9%.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have