<![CDATA[The availability of nitrogen in soil is a crucial factor determining crop productivity. However, the measurement of total nitrogen (N-total) content requires considerable time and cost. Therefore, a fast, accurate, and easy prediction method is needed to support the agricultural development. This study aims to develop an Artificial Neural Network (ANN) model based on the backpropagation algorithm to identify soil N-total content using soil pH, moisture content, and soil resistance as input parameters. The model was trained using the trainbr training function with variations of logsig and tansig activation functions and hidden layer structures of 5–5, 8–8, and 12–12 to obtain the best configuration. The training results indicate that the tansig–tansig combination with 8–8 hidden layer structure achieved the highest performance, with a R2 training of 0.953 and a R2 testing of 0.911. The model was implemented in the form of a Graphical User Interface (GUI) application to facilitate field-level prediction. Validation using 40 testing data samples showed a classification accuracy of 70% and an R² value of 0.932 for nitrogen prediction. The model correctly classified 28 data samples out of the total 40 tested data. These results indicate that the proposed model is capable of predicting soil nitrogen content accurately and reliably.]]>
Copyrights © 2026
