Lokanan Mark
Faculty of Management, Royal Roads University, Victoria, BC, Canada.
Front Big Data. 2022 Oct 10;5:961039. doi: 10.3389/fdata.2022.961039. eCollection 2022.
Investment fraud continues to be a severe problem in the Canadian securities industry. This paper aims to employ machine learning algorithms and artificial neural networks (ANN) to predict investment in Canada. Data for this study comes from cases heard by the Investment Industry Regulatory Organization of Canada (IIROC) between June 2008 and December 2019. In total, 406 cases were collected and coded for further analysis. After data cleaning and pre-processing, a total of 385 cases were coded for further analysis. The machine learning algorithms and artificial neural networks were able to predict investment fraud with very good results. In terms of standardized coefficient, the top five features in predicting fraud are offender experience, retired investors, the amount of money lost, the amount of money invested, and the investors' net worth. Machine learning and artificial intelligence have a pivotal role in regulation because they can identify the risks associated with fraud by learning from the data they ingest to survey past practices and come up with the best possible responses to predict fraud. If used correctly, machine learning in the form of regulatory technology can equip regulators with the tools to take corrective actions and make compliance more efficient to safeguard the markets and protect investors from unethical investment advisors.
投资欺诈在加拿大证券行业仍然是一个严重问题。本文旨在运用机器学习算法和人工神经网络（ANN）来预测加拿大的投资欺诈情况。本研究的数据来自2008年6月至2019年12月期间加拿大投资行业监管组织（IIROC）审理的案件。总共收集了406个案例并进行编码以便进一步分析。经过数据清理和预处理后，共有385个案例被编码用于进一步分析。机器学习算法和人工神经网络能够很好地预测投资欺诈。就标准化系数而言，预测欺诈的前五个特征是违规者经验、退休投资者、损失金额、投资金额和投资者净资产。机器学习和人工智能在监管中具有关键作用，因为它们可以通过从所摄取的数据中学习来识别与欺诈相关的风险，以审视过去的做法并提出最佳应对措施来预测欺诈。如果正确使用，监管技术形式的机器学习可以为监管机构提供采取纠正行动的工具，并提高合规效率，以维护市场并保护投资者免受不道德投资顾问的侵害。
