Penerapan algoritma Long Short-Term Memory untuk prediksi jalan berlubang

Abstrak

The Department of Public Works Bina Marga and Cipta Karya in Central Java Province is a government agency responsible for planning and constructing roads and other public infrastructure. The issue comes when determining which city should receive special focus on road maintenance, considering the road conditions that necessitate more extensive care. The research aims to address the problem by utilizing the Long Short-Term Memory (LSTM) technique to forecast hollow paths. The study utilized data obtained from hollow road recordings collected from 9 Road Management Halls (BPJ) in Central Java Province, spanning from January 1, 2023, to December 31, 2023. The research findings demonstrated that long short-term memory (LSTM) can accurately anticipate hollow pathways on 9 BPJ datasets, as indicated by the Mean Absolute Percentage Error (MAPE) values. Based on the evaluation results, it is evident that the model's performance differs across different locales. The model with the highest level of accuracy is located at BPJ Purwodadi, with a Mean Absolute Percentage Error (MAPE) of 1.932868 percent. This model was trained using a batch size of 48 and 200 epochs. Conversely, the model with the lowest accuracy was observed in BPJ Wonosobo with a Mean Absolute Percentage Error (MAPE) of 30.511073 percent. This model was trained using a batch size of 48 and 50 epochs.