Prediksi Hasil Produksi Tanaman Tomat di Indonesia Menurut Provinsi Menggunakan Algoritma Fletcher-Reeves

Abstract

Tomatoes are essential for Indonesians because they have high economic and nutritional value. In addition, as population growth increases, the demand for tomatoes also increases. Based on this, it is essential to research to predict the future development of tomato crop production. The research in this paper uses a dataset of tomato plant production in Indonesia, which is spread across 34 provinces in the last seven years, namely from 2015 to 2021), which is sourced from the Indonesian Central Bureau of Statistics and the District/City Agriculture Service of each Province. The algorithm proposed in this study is the Fletcher-Reeves Conjugate Gradient Algorithm which will be processed with the help of Matlab2011b. Research analysis with three network architectural models: 5-7-1, 5-13-1, and 5-17-1. Based on a network comparison of the three architectural models, the best result is the 5-17-1 model because the MSE value is the smallest compared to the other two models, namely 0.0009915 compared to 0.0010851 and 0.0049764, as well as the highest level of accuracy, namely by 94% versus 91% and 88%. Therefore the 5-17-1 model is used to predict the yield of tomato production in Indonesia for the future (2022 and 2023). Based on the prediction results at the end of 2022 and 2023, there are 18 provinces where tomato crop production has the potential to increase, although not too significantly. The prediction of tomato production using the Fletcher-Reeves algorithm is quite good because it produces a small error rate and high accuracy.