Using Visual Analytics and Interpretability Strategies to Understand the Impact of Input Variables on Indexes derived from Municipality (Urban) Data Sets
Composite indices have been widely used in several domains as a measure to describe abstract concepts through the combination of variables. The current approaches for index creation and analysis do not have a comprehensive visual interface to enable the use of external information to support interpretation. We propose a visual analytics framework that places users in the loop for creating and interpreting indexes. It helps users to compose an index with the flexibility of determining a weight for the linear combination of indicators. For the interpretation, we use regression analysis to provide explanations for indexes from both internal and external variables. we demonstrated use-case scenarios using crime and demographic datasets to show the benefits of our interface for decision-making tasks at the municipal level. we validate our results through a comprehensive user evaluation, showing that most users reach similar conclusions when using our framework to execute analytical tasks.