Algoritma Deep Learning dalam Memprediksi Hasil Panen Padi di Kabupaten Lamongan

Abstract

Based on data, bps.go.id harvest from the 2nd year 018 to 2019 decreased about an 7.76%. The government must constantly analyze the rice yields of farmers in Indonesia to determine whether these crops can meet the Indonesian people's primary food needs. Research this will predict rice yields in Lamongan. This study aims to assist the government in overcoming the occurrence of significant food shortages in Lamongan. A system that can be used as a reference tool to assist in policy or rule in the district Lamongan. This research proposes deep learning algorithms to predict the harvest based on the land area (m2), spacing (cm), the type of rice, the number of times to fertilize, fertilizer, and crop yields (quintals). The dataset used in the study was collected through questionnaires. Questionnaires were distributed via a google form and contained as many as 390 rows of data. Some of the data produced were incorrect, so the processing was carried out. The results of data processing, the data that can be used are 380 rows. The proposed architectural model's test results show that the loss values of MSE, MAE, or MAPE are the same. The MSE, MAE, and MAPE values are 2939977.418, 301,788, and 83,798, respectively.