The George Boole Foundation
The Decision Analysis Initiative

Reducing financial risks ...

Decision analysis is concerned with the management of uncertainty associated with decisions. In areas such as financial intermediation risks exist not only in terms of market dynamics but also in the fact that decision analysis models applied to the assessment of risk in trading and holding derivatives need to meet high standards of comprehensiveness, coherence and robustness. The recent financial crisis arose as a result of failures to meet such standards.

Expert system diagnostics are of increasing importance in bio-medical applications in assisting in the rapid evaluation of patient circumstances leading to diagnosis and appropriate treatments for cure.

Deduction is a process of inference or arriving at a conclusion as to what a collection of information means. Thus in the case of disease, medical practitioners use information on symptoms to diagnose whether or not a person is ill and, if so, what the cause is. The degree of advance of disease provides additional information used to prescribe an appropriate treatment.

Decision trees

Human mental processes make use of a deductive logic which assess the probability of events and which can be coded as binary logic1. Deductive processes can be automated using a query system to which replies are either yes or no. Thus a sympton exists or it does not. By distinguishing between different states of a symptom a yes or no answer will confirm the status of advance of a condition, and so on. This process makes use of decision trees. Such deductive processes can be applied to areas as diverse as the analysis of environmental exposure and impacts, assessing production potential of agricultural systems, risk assessment in investments and finance, and even litigation analysis.

Inference engines

A decision tree represents a map of the pathways or logical steps required for the deductive process to arrive at reasonable conclusions. It is therefore also a map of the process of inference which can be summarised as an integrated function known as an inference engine. Inference engines are essentially logical processes whereby the many different binary branch responses, together with estimates of observed variable values, are compared to expected distributions of values and, on the basis of this comparison, directed to the next decision tree node through the application of the appropriate decision rules.

Expert dialogue systems

Expert Dialogue Systems
Expert systems can help people follow a sequence of decisions whilst remaining within the realms of feasibility through logical constraints imposed by the state of knowledge of the application. They are often configured as Expert Dialogue Systems (EDS) which consist of a query system structured according to the knowledge base of the domain of the application and the types of decision being undertaken. EDS normally have structured calculators to present quantitative options. Users are guided through a sequence of steps by responding to a sequence of questions (structured queries). The order of queries follows a logical sequence based upon the knowledge structure of a specific domain so as to gradually eliminate the possibility of incorrect decisions in, for example, medical diagnosis. EDS are particularly useful in decision-making in biomedical, financial and litigation analysis.

Applications domains

EDS can be applied in any area including knowledge domains combining complex systems of determinants. It is applied in finance, determination of derivative risk, in management assessment of changing market circumstances, ensuring that all relevant questions have been covered and analysed, biomedical diagnosis including environmental assessments and agricultural production appraisal, crop and animal disease analysis, product indentification and selection from multiple choices and many other applications.

This site is sponsored by SEEL Systems Engineering Economics Labs