Crime Evaluation Using Neural Networks by GPS Trajectory Dataset

One of the many irregularities of the world includes crime. It clumps in some areas and is absent in others. Various daily activities require this knowledge. They seek out particular locations while avoiding others. Their decisions on neighborhoods, schools, stores, roadways, and recreational opportunities are influenced in part by the knowledge that Some of these locations have higher victimization rates than others. These choices further create a reason of developing crime concentrated areas also known as hot spots. Thus, a need to analyze the crime levels and to determine the causes of such concentrations is really useful. Tracking every route of the area with possible reason is only possible by using Global Positioning Systems (GPS) which makes the data availability easier and more accurate on minute level. A study is done this research which finds the best features that might explain the possible criminal activities in a region by building a machine learning neural network (NN) model and minimizing the errors. By the end of this study the top features that came out are ‘accidents’, ‘traffic’, ‘tourist attraction, ‘length of the route’, ‘duration of the route’ which majorly effect the crime rates of an area.