Data Mining Solutions for Sustainability Problems

Nearly every aspect of modern life is laced with questions and choices regarding sustainability. Some questions are pervasive, e.g., should I print this IEEE Potentials article or should I read it online? Others are subtle and we might not think consciously about them, e.g., how much CO₂ does a Google search release into the atmosphere? Still others are knotty conundrums: how do we encourage and incentivize an entire city to “go green?” Computational sustainability [Gomes (2009)] deals with answering questions such as the above using mathematical and algorithmic techniques. Its scope is broad: from designing environmentally friendly substitutes for everyday products, to reducing carbon emissions of data centers, to encouraging energy efficiency in homes, and finally to understanding the interplay between multiple systems at a societal level. Many issues interplay in achieving sustainability goals. First, it is desirable to have an accurate model of the underlying process or product so that we can understand exactly where to focus our sustainability objectives. Second, we must systematically evaluate and assess alternatives alongside multiple (environmental and other) criteria. Finally, satisfactory implementation of sustainable alternatives requires a “buy-in” from all involved stakeholders.