Mining counterfactual antecedents became a valuable tool to mine knowledge and explain machine learning models. It consists of generating synthetic samples from an original sample to achieve the desired outcome in a machine learning model. It can be used to analyze the impact of socioeconomic and urban variables in crime by understanding which changes in a set of variables could alter a region’s definition from unsafe to safe. To help decision-makers interpret crime (and other tasks) define public policies (and other decisions), we propose two methodologies: MAPOCAM and CounterCrime. (1) MAPOCAM is a tree-based search that generates a diverse set of Pareto-optimal counterfactual antecedents by relying on multi-objective concepts. (2) CounterCrime uses various interactive visual metaphors on crime analysis using counterfactual antecedents. We organize our analysis at three levels, whole city, region group, and at the regional level to facilitate exploration. Both methodologies empower users to inspect multiple counterfactual antecedents for the best interventions and have a broader understanding of the prediction.
*Texto informado pelo autor.
Registre-se com antecedência para este seminário aqui.
Após o registro, você receberá um e-mail de confirmação contendo informações sobre conexão no seminário.
Marcos M. Raimundo received the B.S. degree in computer engineering, the M.S. and Ph.D. degree in electrical engineering from the University of Campinas, Campinas, Brazil, in 2011, 2014, and 2018 respectively. He is a post-doctoral researcher at the School of Applied Mathematics at Fundação Getulio Vargas. His research interests include machine learning, multiobjective optimization, mathematical programming, and operations research.