Recommendation System for Automatic Watering and Fertilization of Shallots Using LSTM Algorithm

Modern technologies are currently required for the cultivation of shallots. This research aims to create a smart system that identifies the most effective irrigation and fertilization strategies for shallot cultivation by leveraging the Internet of Things (IoT) and machine learning. The research suggests an automated fertilization method facilitated by advanced technology and explicitly addresses farmers' obstacles, particularly in managing water during dry seasons. The IoT device in this system is equipped with three sensors that are designed to collect data on soil moisture, air temperature, and air humidity. The evaluation results for irrigation classification based on the three sensor conditions show an 83% accuracy when compared to the sole observation of soil moisture. In accordance with the schedule and the height of the plant, fertilization is implemented. The performance evaluation of the Long Short-Term Memory (LSTM) model suggests that it is of high quality, as exemplified by its low Mean Squared Error (MSE) and Root Mean Squared Error (RMSE) values of 0.7 and 0.8 for temperature, 8 and 2 for humidity, and 5 and 2 for soil moisture. The Mean Absolute Percentage Error (MAPE) values of the LSTM model for temperature, atmospheric humidity, and soil moisture are 1.13%, 4.6%, and 3%, respectively.