Game Theory For Decision Support Systems
In this post, I explain a simple game and how Game Theory creates a framework for us to build Decision Support Systems with.
We need a mathematical framework to fit decisions into something we can build models around. Game Theory (GT) provides a solution with limitations.
Decision Support Systems (DSS) are the interfaces between models and people making decisions. What problems do DSSs solve?
We make decisions with incomplete information. The external systems involved in our decision-making processes are too complex for people alone to understand fully. We develop heuristics to cope with limited information.
We introduce bias into our decision-making process that overrides available data. Bias can short-circuit a rational decision-making process turning us into irrational actors and inconsistent decision makers.
Decisions involve consensus. Sometimes we must negotiate with the voices in our head, and in other scenarios, we are working with real people. Collaboration and conflict are part of the process.
Decisions have consequences, both near and long term, which do not have apparent relationships back to their origins. It's hard to improve without a complete picture of the impacts changes have on outcomes. Infrequent decisions are also difficult to improve except over very long timescales.
The purpose of DSSs is to address these main shortcomings. As Data Scientists, we introduce data to the decision-making process and help decision makers discover a better approach for improving outcome quality.
Reporting is an unintentional DSS. Data Scientists must create more rigorous data products, or we may do more harm than good. It is not enough to have data and serve data. Again, people are irrational actors, and our purpose is to introduce a better framework for decision-making.