Determining Factors that Significantly Impact Injury Levels in a Production Facility

Over an eight-month period in 2002, an analysis was performed at Michael D Computer Corporation, to determine how different factors within the production environment at the largest production facility, TMC such as headcount levels, work day, work times, overtime amounts, and percent of temporary employees impacted productivity. Using a linear regression model, a prediction equation was created to predict the output in the factory given the above listed factors. While the work was informative, and productivity could be modeled, these factors also impact other non-productivity outcomes, such as product quality defect levels, as well as the number of injuries seen on the factory floor. While the initial publication included models and results that could be used to predict the number of quality defects using the factors listed above, determining the number of injuries was not determined at that time. This is because modeling of injuries in the factory is much more complex because it relies on the use of more complex non-linear regression techniques. This paper uses the production information from the TMC factory, and determines which factors are important in predicting the number of injuries that will exist in the factory by using an advance nonlinear regression methodology, known as the Poisson Loss Function. Through the use of this technique, it could be determined, that the major contributors to Injuries in the factory, were the Total Headcount in the factory, the Percentage of Temporary Headcount utilized in the factory, the Shift being worked, and the Day of the Week being worked.