A Real Time Malfeasance Prediction for Multivariate Spatio-Temporal Data Using Novel Geospatial Algorithm Compared with Reinforcement Learning Algorithm

In order to forecast the Malfeasance Prediction for Multivariate Spatio-Temporal Data in a real time environment in urban and rural areas, this work compares the Novel Geospatial Algorithm with Reinforcement Learning Algorithm. The dataset used in this article comprises Data sources (Crime reports and Incident Data, Geospatial Data). Novel Geospatial Algorithm and Reinforcement Learning Algorithm are simulated by various parameters and automate Malfeasance prediction in urban and rural areas. The sample size for each group is 20 which is determined by Gpower test, and the Gpower setting parameters were <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\alpha=0.05$</tex> and power <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$=0.85$</tex> Based on obtained results Novel Geospatial Algorithm has significantly better accuracy (93.3400) compared to Reinforcement Learning Algorithm accuracy (85.9260). Statistical significance difference between Novel Geospatial Algorithm and Reinforcement Learning Algorithm was found to be <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$p=0.276$</tex> (Independent Sample T Test <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$p&gt;0.05$</tex>). This study shows that the Novel Geospatial Algorithm provides better efficiency results in predicting the Malfeasance compared to the Reinforcement Learning Algorithm.