Casualty reduction intelligent system based on classified prediction and comparative analysis of industrial mishaps

Abstract

Industrial accident analysis is a very challenging task and one of the most vital issues in the era of globalization. Discovering the attributes becomes more complex because voluminous factors are associated. We have tried to find out the specific attributes and made a cumulative dataset depending on the reliable sources allied to Bangladesh. In the study, we have evaluated a meticulous survey on various classification techniques to achieve casualty for textile & garments accidents. We have presented a comparative analysis of accuracy between base and AdaBoost Meta classifier using base classifiers, such as: OneR, J48, REPTree, SimpleCART & Naïve Bayes. The analysis unfurl that using ensemble method with the base classifiers improve accuracy level between 1.8%-6.36%. We have also proposed a system named “Casualty Reduction Intelligent System (CRIS)” depending on the knowledge explored by classification technique which will have the ability to make automated decision that is quite similar to human decision making for reducing the rate of casualty of industrial mishaps.