The Land use management constitutes a multi-dimensional issue affected by a variety of criteria of different significance. Many decision-makers (DMs) are involved in this type of dilemma, and their preferences are often in dispute. To address these issues, researchers created a variety of GDSS with various architectures; nevertheless, not all of them can apply artificial intelligence approaches to mimic human behavior by predciting or classifying solutions. In this work, the authors used a previously designed GDSS named WIM-GDSS as the foundation for developing a new one with various features; the two systems differ in the prediction model employed. The proposed system’s prediction module employs a model trained on a multicriteria method known as PROMETHEE II rather than TOPSIS; the latter method is widely used in the literature and provides more choice and flexibility to the user when expressing preferences (more subjective parameters than TOPSIS). The paper includes a real case study in territorial planning, in which the proposed system would manage a group decision-making process for selecting the most suitable vacant zones for housing building. A coordination protocol will ensure DMs cooperation. The AHP approach will be used to assign criteria weights based on the preferences of DMs. This system includes a prediction module that predicts solutions rather than calculating them using a prediction model. In order to choose the optimal model, a comparison study was done between two models: Linear Regression (LR) and Multi Layer Perceptron (MLP). The results suggest that the MLP model is more suited to PROMETHEE II than the LR model, with a 95% accuracy. Future study will broaden the trials to include fuzzy logic approaches and completely integrate the proposed system with the geographic information system.