Spatiotemporal prediction framework for detecting illegal online transactions in the digital economy

The rapid growth of the digital economy has transformed online transaction channels, payment methods, and supply chains, but it has also increased illegal activities, challenging the supervision and security of online transactions. Predicting illegal online transactions is crucial for effective cross-regional prevention and control. To address this, we propose a spatiotemporal prediction framework. The CEEMDAN algorithm decomposes time series data to capture periodic and trend changes, while a topology map models spatial correlations. Combining Spatio-temporal Graph Convolutional Network (STGCN) and Long Short-Term Memory (LSTM), the model identifies high-risk areas and extracts key transaction behavior factors. Experimental results demonstrate the model’s strong performance in predicting illegal online transactions.