Predictive framework for crime data analysis using a hybrid logistic regression - support vector machine based ensemble classifier powered by CART (LR-SVMCART)

Significant rise in illegal activity has directly impacted socioeconomic growth and quality of life. In this article, a predictive crime data analysis framework has been proposed that can resolve the problem of scalability issues and accuracy rate. This paper proposed a hybrid ensemble machine learning classifier to identify authentic crime activities. A series of experiments are used to verify the efficiency of our proposed algorithms. Three datasets of different countries are used for this experiment purpose. All the datasets are tested successfully on our proposed framework and novel ensembles classifier. The result produced by our proposed hybrid ensemble classifier mostly outperforms the performance of most of the existing machine learning approaches. This work aims to identify geospatial crime data intensity where we can anticipate the recurrence of a certain crime in the city using geospatial technology, allowing the police force to take the required precautions to avoid it.