Abstract
The high demand for chilies, especially in Gorontalo, is a driving force for chilli cultivating farmers. The price of chili which is uncertain every day can fluctuate. The Gorontalo City Food Service cannot make predictions to estimate prices in the following month. Prediction is defined as the use of statistical techniques in the form of a picture of the future based on the processing of historical figures. Due to the many algorithms that can be used in predictions, this study will compare forecasting algorithms namely Support Vector Machine (SVM), K-Nearest Neighbor (K-NN) and Neural Network (NN). Experiments that have been carried out, on chili price prediction with forecasting algorithms have been successfully carried out. The root mean square error (RMSE) result of the SVM algorithm is 0.233, the K-NN algorithm is 0.223 and the NN algorithm is 0.206. Of the three forecasting algorithms used, the best results are produced by the Neural Network algorithm with the smallest RMSE value of 0.206. So it can be concluded that the proposed model is close to perfection, because a comparison of the results of implementing chili price predictions for the next three months produces an accuracy value of 99.25% on average
