Multi-Level KPIs Prediction using Machine Learning for Mining-Site Production Management and Planning

点击翻译按钮获取中文摘要

Measuring and improving performance depends on the adequate selection of Key Performance Indicators (KPIs). Most industrial firms, including Goldcorp rely on various KPI systems for balancing financial (production) and non-financial (management and planning) measures in achieving their strategic goals. In order to develop such KPI systems, experts usually perform in-depth evaluation of the company's business goals using a range of quantitative and qualitative statistical measures. The latter are usually derived from a large number of organizational, operational and business-model features. The limitations of this approach, in predicting KPIs, that they are mainly dependent on the quality and the amount of data used in the analysis, the statistical methods used, and the level of experience of those conducting such business analytics. Today, with the advances in artificial intelligence, machines are capable to imitate intelligent human behavior in solving a variety of recognition and prediction problems relying on big data analytics. The aim of this project is to assist Goldcorp in applying state of the art deep learning algorithms in order to discover new insights about the best KPI pattern, thereby enabling the company to optimize its production and planning, while optimizing the use of its resources. The benefit of this project venture is the training of highly qualified employees for the Canadian mining industry, which increasingly requires highly specialized expertise in Artificial Intelligence and Big Data Analytics to develop intelligent production systems. Building local competencies and having access to well-trained graduates who are proficient in such advanced IT fields is a major source of competitiveness for Goldcorp and Canadian mining industry, and it is expected to contribute to job creation.