CriPaaP: A Geospatial Crime Pattern Analysis and Prediction Framework Integrating DBSCAN, Enhanced LSTM, and ST-GNN for Urban Safety in Nigeria | IJCT Volume 12 – Issue 5 | IJCT-V12I5P57

Urban crime presents significant challenges to law enforcement and community safety. This study introduces the Crime Pattern Analysis and Prediction (CriiPaaP) model, a hybrid geospatial forecasting framework integrating DBSCAN clustering, an enhanced Long Short-Term Memory network (eLSTM), and a Spatial-Temporal Graph Neural Network (ST-GNN). The model leverages spatio-temporal-contextual data to predict crime patterns at identified hotspots. DBSCAN first detects geospatial crime hotspots, which are then encoded as graph nodes. Temporal dynamics are captured using eLSTM with attention and batch normalization, while ST-GNN encodes spatial dependencies through graph convolutions. Predictions from both models are fused at the output layer. Results demonstrate that eLSTM achieved the lowest RMSE (0.032), while the fusion model provided balanced forecasts (RMSE 0.055) with stable performance across clusters. The study shows that combining spatial and temporal learning yields more reliable hotspot forecasting for law enforcement resource allocation.

This study introduced the Crime Pattern Analysis and Prediction (CriiPaaP) framework, which is a fresh hybrid model that combines DBSCAN clustering, an enhanced LSTM (eLSTM), and a Spatial-Temporal Graph Neural Network (ST-GNN) to tackle the hurdles of geospatial crime forecasting in Nigeria. By merging spatial hotspot detection, learning from temporal sequences, and utilizing graph-based spatial-temporal modeling, the framework offers a thorough approach to predicting not just where crimes might happen, but also when they’re likely to occur. The findings showed that the eLSTM performed particularly well in terms of temporal accuracy, while the ST-GNN effectively captured important spatial relationships. The Fusion model, which brings both of these elements together, provided the most reliable and easy-to-interpret predictions, as indicated by the forecast plots and distribution of residuals. For law enforcement, this dual capability gives agencies useful insights that aid in planning proactive patrols, allocating resources strategically, and enhancing community-driven crime prevention efforts. Looking ahead, there should be an emphasis on enhancing CriiPaaP with real-time mobility data—like transportation patterns and mobile phone usage—as well as socio-economic factors to make predictions more context-aware. Furthermore, validating the model in various cities across Nigeria, along with comparing it to international examples, will be essential for determining its scalability and applicability. By aligning innovative methods with real-world usage, the CriiPaaP framework signals progress toward data-informed, preventive policing strategies in rapidly expanding urban areas

Adedamola, A. O., & Ogundunmade, T. P. (2025). Predictive Modelling of Crime Data using Machine Learning Models: A Case Study of Oyo State, Nigeria. International Journal of Innovative Science and Research Technology, 10(4), 1669-1677. Chainey, S., & Ratcliffe, J. (2013). GIS and crime mapping. John Wiley & Sons. Du, Z., Liu, Y., Zhang, W., & Li, R. (2016). Crime trend prediction using LSTM. In 2016 International Conference on Security, Pattern Analysis, and Cybernetics (SPAC) (pp. 57-61). IEEE. Ester, M., Kriegel, H.-P., Sander, J., & Xu, X. (1996). A density-based algorithm for discovering clusters in large spatial databases with noise. In Proceedings of the Second International Conference on Knowledge Discovery and Data Mining (KDD-96) (pp. 226-231). Oladele Robert, B. (2025). Development of a predictive model for crime investigation and emergency response system. Federal University of Technology Owerri (FUTO) Repository. Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, Ł., & Polosukhin, I. (2017). Attention is all you need. In Advances in Neural Information Processing Systems 30 (NIPS 2017) (pp. 5998-6008). Wu, Z., Pan, S., Chen, F., Long, G., Jiang, J., & Zhang, C. (2020). A comprehensive survey on graph neural networks. IEEE Transactions on Neural Networks and Learning Systems, 32(1), 4-24. Yu, Y., Yin, H., & Zhu, Z. (2018). Spatio-temporal graph convolutional networks: A deep learning framework for traffic forecasting. In Proceedings of the 27th International Joint Conference on Artificial Intelligence (IJCAI-18) (pp. 3634-3640). Zhao, P., Liu, T., He, S., & Li, Y. (2021). Spatio-temporal crime forecasting with graph neural networks. Journal of Big Data, 8(1), 1-22.

Call for Paper (Hub) Fast Paper Publication IJCT Indexing & Databases Submit Manuscript Call for Paper: Sep–Oct 2025 Call for Paper: Publish Research