Time series prediction of economic indicators in Indonesia using differential dynamic optimized by genetic algorithm

Prediction time series of economic indicator optimized by Algorithm Genetic (AG) is able in getting best individual with the accuracy around 97%. The parameter of AG are maximum population is 100; Ra are 5 and 10, while Rb are -5 and -10; probability mutation (Pmut) is 0.3; and probability crossover (Pc) is 0.9. It was caused by AG had longer opportunity in fitting data using scenario data from 1961 until 2012 than other scenario. Besides, it came from using differential dynamic that can solve the chaos and complex of economic indicator. Using other parameter AG which combined with few data training or testing will affect much in declining the accuracy around 20%. It associated with learning in AG that done really fast, while the function fitness in getting best individual not yet matches with the optimization predicted. Because of that, AG needs equation that can suitable with the whole scenario of data which will be used.