資源経済

[1K0501-05-03] Unsupervised machine learning for Critical Minerals identification

○Malala Ojiambo¹, Tsuyoshi Adachi¹ (1. Akita University)

Keywords:Critical Minerals, Machine learning, Supply risk, Long-term criticality

Worries about the supply of minerals resources have re-emerged in recent years. Unreliable and interrupted supply of mineral resources may have severe repercussions because industries and modern technologies depend on these resources. The concerns have led researchers to propose ways to filter the most critical mineral resources- minerals that have high economic importance and face high supply risks- as a mechanism to preempt supply disruptions and induce countermeasures. However, in identifying critical minerals, the studies use different indicators chosen arbitrarily and aggregated with different techniques leading to conflicting results and interpretations. This study explores the use of unsupervised machine learning techniques to identify long-term critical minerals. Machine learning techniques can, with minimum human interference, analyze patterns from vast amounts of data such as production and prices and filter those minerals that have a high supply risk yet crucial to industry and modern technology. The study aims to create a verifiable tool for governments and companies to identify critical minerals. The study is ongoing, and the initial results will be presented and the implications discussed. Our study is essential because it can catalyze decisions to reduce dependence on particular minerals, promote efficiency in using of some materials, and help prioritize the research agenda.