Daniel J. Fonseca

An expert system for financial ratio analysis

Financial analysis interprets a companys past and present financial health and predicts its future condition. Although company financial statements contain a wealth of information to support this analysis, their interpretation may be complicated. Experts in this field are limited. This research focuses on automating the current practice of financial ratio analysis to identify the various features that need to be incorporated into the system. This involves calculating the ratios, establishing the relationships between the ratios, determining the technique for accurately forecasting the financial statements and/or ratios, developing heuristics for analysing the ratios and providing a system for recommendations. A prototype expert system was then developed. The system is capable of performing five types of analysis: liquidity, leverage, turnover, profitability, and past performance. The output of the system is a list of conclusions and recommendations based on these analyses.

Evacuation Decision Support System for Road Incident Detection and Characterization

This research effort focused on the development of an automatic road incident detection and characterization system which will lead to the reduction of non-recurrent traffic congestion on freeways. An algorithmic methodology was developed for incorporation and enhancement of the existing South Carolina Department of Transportation (SCDOT) evacuation analysis system. Analysis of the existing systems input data helped determine the specific model needed. Investigation included a series of traffic analysis algorithms that consider: 1) identification of prevalent traffic flow conditions during a predetermined time window, 2) recognition of incident occurrence, 3) incident characterization, and 4) subsequent routing. The algorithmic methodology initially developed by Sheu (2002) was considerably expanded and adapted to the available SCDOT traffic data. This modified approach was then incorporated into a general system design document to establish definitions and descriptions of the proposed enhanced system. This initiative will provide traffic officials with relevant insights and expertise to detect and characterize emergency situations. It represents a significant improvement over the labor-intensive traffic monitoring systems currently used by many state DOTs, which lack automated capabilities for incident detection and characterization.

The Role of Non-Recurring Congestion in Massive Hurricane Evacuation Events

The response to a potential disaster can require the evacuation of personnel from a specified area. Generally, such efforts are restricted to the orderly mass departure of individuals across pre-planned and well maintained transportation routes. In the U.S., evacuations of up to 1,000 subjects take place every two to three weeks, with more extreme evacuations involving two million or more every one to three years (TRB, 2008). While evacuation routes are designed to accommodate normal traffic movements, congestion and gridlock can occur as the design capacity of the road system is overwhelmed by the magnitude of vehicles leaving the affected area. The resulting traffic patterns affect the safety and mobility of subjects moving to more secure areas. Adding to this disarray, potential nonrecurring incidents congest traffic patterns even more. Estimates indicate that between fifty and sixty-five percent of traffic congestion is caused by non-recurring traffic incidents with an additional ten percent related to construction and weather (Coifman, 2007). A non-recurring traffic incident is any event that both causes a reduction of roadway capacity, or an abnormal increase in demand, and requires first responders to be dispatched. Stalled vehicles, roadway debris, spilled loads, and crashes fall into this category of incidents. Non-recurring traffic incidents can cause secondary traffic incidents. These incidents further congest the traffic stream and cause delays in clean-up efforts by first-responders. Studies indicate that twenty percent of traffic incidents are secondary incidents, with one out of five resulting in a fatality. In addition to crashes, secondary incidents can include overheated vehicles, out of fuel conditions, and engine stalls. The delay and traffic gridlock associated with traffic incidents is compounded during the evacuation process due to the large numbers of subjects leaving the affected area. These delays and backups result in: • Increased response time by first responders • Lost time resulting in a wider evacuation window • Increased fuel consumption • Reduced air quality and other adverse environmental conditions • Increased potential for more serious secondary incidents resulting from rear end collisions, traffic exiting the route, or exiting to the shoulder of the road • Increased potential of crashes by incidents involving personnel responding to traffic incidents • Negative public image of first responders involved in incident management activities.

A hybrid system approach for space station risk analysis

This paper describes a conceptual design for an integrated hybrid system for risk analysis applications on the International Space Station project. The expert systems knowledge base incorporates heuristics to integrate and interpret the algorithmic results from both simulation (for project risk analysis) and probabilistic risk assessment (regarding safety issues). This project was conducted in four phases: model base analysis, knowledge engineering, knowledge base design, and development of the functional system design. A functional system design document was developed in order to identify the detailed system architecture, processing considerations, and definition of displays and reports. Recommendation of software products was based on investigation of functionality, installation, and integration characteristics, as well as compatibility with existing NASA hardware, software, and communications investments. A prototype system was then developed.