An Interpretable Deep Learning Framework for Assessing Financial Potential of Urban Spaces

In this work, we propose a novel deep learning framework to predict the future financial potential of urban spaces. We use the number of financial institutions as our prediction target in an urban area. Our model offers three kinds of interpretability, providing a better way for decision makers to understand the decision processes of the model: a) critical rules that determine the prediction; b) influential surrounding grids; and c) critical regional features. Our module takes advantage of a tree-based model, which can effectively extract cross features. Our proposed model also leverages convolutional neural networks to obtain more complex and inclusive features around the target area. Experimental results on real-world datasets demonstrate the superiority of our proposed model against the existing state-of-art methods.