Figure 4

Model Validation Confusion Matrix: (a) Artificial Neural Network (ANN); (b) Gradient Boosting Decision Trees (GBDT); (c) Logistic Regression (LR); (d) Self-Attention Residual Neural Network Diagram (SARes-NET); (e) Naive Bayes (NB); (f) Random Forest (RF). Note that we use ‘0’ as the label for Ecology; ‘1’ as the Agriculture label; ‘2’ as the label for Urban. Each row of the matrix corresponds to a predicted lable, including ecological, agricultural, and urban spaces. Each column of the matrix corresponds to an true lable. The values on the diagonal represent the number of consistent predictions with the ground truth for each category. Compared to other models, our model has the highest number of correct predictions, indicating superior performance of the matrix.