Logical analysis of data for ranking green technologies

Abstract

Selection of the best green technology for a given application is a multi-criterion decision analysis problem. The tradeoffs that need to be made among conflicting attributes of the alternatives are inherently subjective and require the use of interactive decision analysis tools. In this work, a machine-learning based methodology is developed for ranking green technologies based on multiple criteria. First, an expert generates a training data set by ranking a subset of the alternative technologies using his/her tacit knowledge and preferences. Then, a machine learning procedure known as logical analysis of data (LAD) is used to detect patterns in these rankings. These patterns are approximations of the mental rules used by the expert to compare and rank the competing alternatives; after validation, they can be used to rank other technologies in the same class as those in the training data. This novel methodology is illustrated here for the case of ranking alternative energy storage technologies for stationary applications. This technique provides a rapid means of eliciting expert knowledge for ranking alternatives based on multiple criteria.

Keywords

• Technology selection
• Multiple-attribute decision-making
• Machine learning
• Rule induction
• Pattern recognition
• Renewable energy