Caner Hamarat

An exploratory approach for adaptive policymaking by using multi-objective robust optimization

Abstract Developing robust policies for complex systems is a profound challenge because of their nonlinear and unpredictable nature. Dealing with these characteristics requires innovative approaches. A possible approach is to design policies that can be adapted over time in response to how the future unfolds. An essential part of adaptive policymaking is specifying under what conditions, and in which way, to adapt the policy. The performance of an adaptive policy is critically dependent on this: if the policy is adapted too late or too early, significant deterioration in policy performance can be incurred. An additional complicating factor is that in almost any policy problem, a multiplicity of divergent and potentially conflicting objectives has to be considered. In this paper we tackle both problems simultaneously through the use of multi-objective robust simulation optimization. Robust optimization helps in specifying appropriate conditions for adapting a policy, by identifying conditions that produce satisfactory results across a large ensemble of scenarios. Multi-objective optimization helps in identifying such conditions for a set of criteria, and providing insights into the tradeoffs between these criteria. Simulation is used for evaluating policy performance. This approach results in the identification of multiple alternative conditions under which to adapt a policy, rather than a single set of conditions. This creates the possibility of an informed policy debate on trade-offs. The approach is illustrated through a case study on designing a robust policy for supporting the transition toward renewable energy systems in the European Union. The results indicate that the proposed approach can be efficiently used for developing policy suggestions and for improving decision support for policymakers. By extension, it is possible to apply this methodology in dynamically complex and deeply uncertain systems such as public health, financial systems, transportation, and housing.

A genetic algorithm based feature weighting methodology

In this paper a genetic algorithm based feature weighting methodology that is based on k-nn classifier is presented. The performance of the algorithm is evaluated in two folds. First of all, its differentiation capability among relevant and irrelevant features is evaluated. This is achieved by introducing dummy variables to a well known benchmark data set, namely the Iris Data. Secondly, its predictive performance is also evaluated. The results are encouraging in the sense that the proposed algorithm specifies lower weights to the dummy variables and yields high classification accuracy.

A Decision Support System framework for innovation management

In this paper, we present a Decision Support System (DSS) framework for innovation management. The proposed framework utilizes the data acquired by a web based questionnaire which is filled in by the upper managers of the manufacturing companies. The DSS compares their answers with the responds of the other companies existing in the database of the DSS and determines the weak points of the company which have room for further improvement and at the same time significantly affect innovativeness. Finally, the DSS automatically generates two reports. The first report benchmarks the company with the others and the second report suggests policies that might be helpful to improve the innovativeness of the company.