Policy Clusters: Government's Agenda Across Policies and Time

In the last decade, Machine Learning research has developed several data analysis algorithms for real-world problems. On the other hand, analyzing the attention governments allocate to different policy areas is important since it helps us to understand the extent to which the limited resources of governments are focused or diversified. We classify the previous studies on government agenda representation into Individual and Total approaches. While the Individual approaches focuse on one policy area at a time and traces the extent of attention each one received, the Total approaches propose aggregated data analysis methods to represent the government agenda considering all the policy areas. In this paper, we use hierarchical clustering to propose an intermediate type of policy analysis called "Policy Cluster" which considers the relationships among different policy areas. For the evaluation, we built and analysed the Policy Clusters for the Irish government covering the time period 1945 to 2012. Comparing to previous Individual and Total approaches, the proposed intermediate approach reduces the search space in which we are looking for informative patterns by 57% and the results of our analysis represent the political agenda in more modular and informative way, taking into account intra-relationships of policies.