Multivariate Malfeasance Prediction in Spatio-Temporal Data Using a Novel Geospatial Algorithm: A Comparison of Accuracy with Anomaly Detection Algorithms

To predict malfeasance in multivariate spatiotemporal data within both rural and urban environments, this study compares the effectiveness of a novel geospatial algorithm against an anomaly detection algorithm. The dataset used comprises various data sources, including crime reports, incident data, and geospatial information, all sourced from Kaggle (2022). The simulations of both algorithms were conducted with various parameters to enhance the automation of malfeasance prediction in these settings. Each group consisted of a sample size of 20, determined using a GPower test with settings of α=0.05 and power = 0.85. The results indicate that the novel geospatial algorithm achieved a significantly higher accuracy of 94.21% compared to the anomaly detection algorithm's accuracy of 81.03%. However, the statistical analysis revealed a p-value of 0.278 from the independent sample t-test (p > 0.05), indicating that the difference in performance is not statistically significant. Overall, this study demonstrates that the novel geospatial algorithm is more efficient in predicting malfeasance compared to the anomaly detection algorithm.