Today: Sunday, 22 December 2024
-
Information Technology
-
Computer
-
Telecommunications
An Ensemble Learning Approch For Crime Analysis And Detection
Volume 2, Issue 2, 2020, Pages 31 - 41
1 Assistant Professor of Computer Department, West Tehran Branch, Islamic Azad University, Tehran, Iran
2 Master Degree, West Tehran Branch, Islamic Azad University, Tehran, Iran
Abstract :
The rate of crimes has substantially increased with the passage of time, and technologies that have helped people enjoy an easier life have contributed criminals to employ more accurate techniques in committing crimes. One of the key concerns for law enforcement officials is how to enhance the police investigative efficacy and attempt to improve it so that they could remain in the eternal contest between law enforcers and criminals. Data mining is the strongest and best method to extract basic knowledge and relationships between data and detect patterns in large amounts of data using various sciences. As a result, crime prediction, crime prevention, and crime detection, scientifically not just empirically with the help of data mining, is a strategy that causes the adoption of better decisions and strategic planning at the micro and macro levels. In this paper, with the help of data mining algorithms and CRISP (Cross Industry Standard Process for Data Mining) methodology, we have dealt with the intelligent identification of criminals. Given that each of the data mining techniques and algorithms has different advantages and disadvantages, the use of ensemble methods that the jury actually constitutes and announces the final decision by a maximum voting will lead to attaining the best result. Our proposed model was trained by ensemble learning classifiers methods from three based classifier algorithms of RandomForest, Naive Bayes, and support vector machine (SVM) with weights of 2, 1 and 4, respectively. This technique has a greater efficiency than other methods and base algorithms and increases the evaluation criteria for the precision and accuracy of classification to 74% and 75%.
The rate of crimes has substantially increased with the passage of time, and technologies that have helped people enjoy an easier life have contributed criminals to employ more accurate techniques in committing crimes. One of the key concerns for law enforcement officials is how to enhance the police investigative efficacy and attempt to improve it so that they could remain in the eternal contest between law enforcers and criminals. Data mining is the strongest and best method to extract basic knowledge and relationships between data and detect patterns in large amounts of data using various sciences. As a result, crime prediction, crime prevention, and crime detection, scientifically not just empirically with the help of data mining, is a strategy that causes the adoption of better decisions and strategic planning at the micro and macro levels. In this paper, with the help of data mining algorithms and CRISP (Cross Industry Standard Process for Data Mining) methodology, we have dealt with the intelligent identification of criminals. Given that each of the data mining techniques and algorithms has different advantages and disadvantages, the use of ensemble methods that the jury actually constitutes and announces the final decision by a maximum voting will lead to attaining the best result. Our proposed model was trained by ensemble learning classifiers methods from three based classifier algorithms of RandomForest, Naive Bayes, and support vector machine (SVM) with weights of 2, 1 and 4, respectively. This technique has a greater efficiency than other methods and base algorithms and increases the evaluation criteria for the precision and accuracy of classification to 74% and 75%.
Keywords :
Intelligent Crimnials Identifcation, Data Mining, Crime, Classification, Ensemble method
Intelligent Crimnials Identifcation, Data Mining, Crime, Classification, Ensemble method