<p>Digital financial inclusion (DFI) plays a pivotal role in alleviating information asymmetry and mitigating financial risk. Accurate prediction of DFI trends is crucial for government agencies and financial institutions to promote financial inclusion and strengthen financial reform and innovation. However, existing studies often overlook regional correlations and dependencies in predicting DFI. We propose a novel approach named Exploratory Spatial Data Analysis and Graph Convolutional Neural Network (ESDAGCN), which integrates Exploratory Spatial Data Analysis (ESDA) and Graph Convolutional Neural Network (GCN). The methodology consists of three main steps. Firstly, we enhance the adjacency matrix construction in GCN by incorporating global Moran’s I analysis to capture the financial background knowledge of DFI and improve the interpretability of the graph. Secondly, we employ local Moran’s I to quantify spatial dependencies among neighboring regions and construct the adjacency matrix based on the resulting Moran matrix. Finally, we design the ESDAGCN model, integrating ESDA and GCN, to predict DFI trends. Experimental results on a dataset focusing on DFI in China demonstrate that the proposed ESDAGCN approach outperforms other baseline methods in terms of predictive accuracy and performance. By incorporating spatial analysis into the GCN framework, the ESDAGCN model captures regional correlations and dependencies in DFI trends prediction, improving both prediction accuracy and model interpretability. This approach provides valuable insights for policymakers and financial institutions in their decision-making processes. </p>