The modelling of business rules for dashboard reporting using mutual information

The role of a business or military dashboard is to form a succinct picture of the key performance indicators that govern an organisation's dynamics or effectiveness. These indicators are based on the judgments of different experts. Variables from one or several databases along with the subjective opinion of the expert are fused or amalgamated to form the so called key performance indicators. A principal example of such a dashboard is the reporting that occurs in the defence preparedness system. Judgments, based on data or opinions within lower level units are formed then fused through the defence hierarchy. The aim of this system is to form a concise picture of the strategically important preparedness issues pertinent to senior leadership. The issue that we begin to address in this paper is the analysis of just how informative such business dashboards can be. There are three principal reasons as to why the information conveyed by the dashboard to senior boards or defence leadership may be uninformative. The first is that the selection of the key performance variables may not capture the true dynamics of the organisation's state over time. The second is that the models or algorithms used to map fundamental database variables onto key performance indicators may be wrong. Finally, even if the models are correct, the key performance indicator may not convey sufficient information as to the state of the organisation. In this paper we will discuss the last issue-that of the information conveyed through the application of business rules to form the key performance indicator. A number of examples will suffice to illustrate the loss of information. Important information might be lost when variables are fused or amalgamated due to the requirement for brevity. Take for example overall profit. While the profit for an organisation or division may be good, the successes of some branches may hide a critical loss in others. A simpler example is the requirement for hierarchical reporting. Reports are amalgamated at the branch level and then to the division level with inevitable loss of information. For this paper we apply the information-theoretic concept of mutual information between the performance indicators -the state-at one level of the organisational hierarchy and the corresponding state, formed by the business rule for fusion at a higher level. The analysis is done for a number of business rules. The business rules analysed are the commonly applied "report by exception" rule or the "report by majority" rule. By calculation of the mutual information, one is able to quantify just how much information is lost through the application of such rules, as reports propagate upward in the hierarchy. We draw some conclusions as to which business rule is more appropriate in organisations that are either relatively static with few key performance indicators changes, versus organisations which are highly dynamic, where such indicators may change often. A generalised business rule is then defined. This work forms the basis for the measurement of the effectiveness of the reporting system itself. Measures of system bias are formed that suggest the degree of effectiveness of the overall system in the communication of the defence system's overall state to the apex of the hierarchy, that is to senior leadership.