Artificial Intelligence–Based Risk Analysis for Management System Certification: A Predictive Application for Moroccan Enterprises

Abstract

A management system is a set of processes that an organization uses to ensure that it can meet its objectives. Certification to international management system standards (ISO 9001, ISO 31000 …) is becoming important for companies that want to demonstrate their commitment to requirements. However, to achieve and maintain certification, companies must undertake a risk analysis process that identifies and ensures that controls are in place to manage them. Companies have started to use technologies to help with risk analysis. Artificial Intelligence (AI) tools like deep learning and machine learning have rapidly evolved recently to improve their decision-making processes. It allows computers to learn and improve from experience, using algorithms inspired by the human brain. The real power of these tools comes from their ability to learn from vast amounts of data, often in an unsupervised manner: the algorithms can identify patterns and relationships within the data, without being explicitly programmed to do so. In this article we will present our results for first application used to predict risks’ criticality related to companies’ activities, gathering data from a database of 29560, that could be used to feed the risks analysis for Management System, in compliance with Standards requirements for the certification, that serves as a Decision Support System (DSS). The proposed FNN, MLP and XGBoost models achieved an R² of 0.9991, 0.9986 and 0.9996 and a Mean Absolute Error (MAE) of 2.93, 3.54 and 2.31, outperforming other models in predicting risk criticality.