The Prediction Analysis Laboratory @ MIT (also called the "Predictics Lab") is led by Professor Cynthia Rudin. We design predictive models and knowledge discovery systems that can more directly assist decision makers than current methods. Our expertise is in machine learning, data mining and knowledge discovery problems. Current research areas include:
◊ Interpretable Predictive Modeling: Designing models that are interpretable to human experts, yet are optimized over large datasets for accuracy. These methods are ideal, for instance, for reverse-engineering better medical scoring systems.
◊ Combining Machine Learning with Decision Making: Envisioning and building predictive systems that know in advance how they are going to be used in decision making, and adapt intelligently.
◊ Learning to Rank: Modeling how entities should be prioritized. These methods involve optimizing rank statistics.
◊ Public Good: We work in many application areas, including several areas of critical importance to our society. These include:
◊ Energy grid reliability: We work on statistical modeling and knowledge discovery problems related to energy grid failure prediction in underground distribution networks in cities.Contact Information
◊ Healthcare: We have been building interpretable predictive models for predicting medical conditions.
◊ Computational Criminology / Predictive Policing: We have been building algorithms that locate crime series committed by a single individual from within a large database of crimes.
◊ Information Retrieval: The goal is to build the next generation of search engines, that not only locates where to find the information, but reads the webpage and brings it back for you. We have been focusing on the problem of "set completion".
◊ Analysis of Meetings: 11 million meetings occur every workday in the United States. We have been taking a quantitative approach to the study of meetings.
◊ Product Quality Rankings: Product quality rankings are supposed to measure quality of a product, but manufacturers cannot optimize the quality of their products without knowing how the rating systems are constructed. We reverse-engineer rating systems from data.
100 Main Street
Cambridge, MA 02142, USA