<![CDATA[Curly chili is a basic necessity for the people of Malang City, namely as a complement to cooking spices so that its existence is often sought after. This causes fluctuation due to the influence of the amount of demand on price change. So a price prediction system for curly chilies is needed in Malang City to minimize price instability. Extreme Learning Machine (ELM) is a prediction method that has high accuracy and faster execution time. ELM does not have a feature selection function, so an optimization method such as Particle Swarm Optimization (PSO) is needed. PSO is implemented as a solution to get optimal weight with the fitness value as a comparison. Based on the tests that have been carried out on the price of curly chilies, the average MAPE value is 1,133803% and the average fitness value is 0,400346 with optimal parameters consisting of 2 features, hidden neuron is 3, the percentage comparison between training and testing data is 90%: 10%, the weight of inertia is 0,5, c1 is 3, c2 is 1,5, the lower speed limit value is -0,8, the speed upper limit value is 0,8, the population is 100, and it is carried out by 260 iterations. From the test results, it can be concluded that PSO is able to optimize the ELM weight so it could gets optimal accuracy.]]>
Copyrights © 2021
