Regression analysis the way of the future, January 13, 2008, 2D.

Imagine the advantages of being able to see into the future and accurately predict things that will happen.  Students would be able to know exactly what to study and learn for assignments and exams, parents would know how to help their children avoid harmful situations, investors could pick rewarding opportunities and avoid costly ones, and emergency workers could head off hazardous and harmful situations before they arise.  The ability to accurately see the future could help people make decisions in the present that lead to things they want in the future.

Organizations and businesses of all types could also benefit from knowledge of the future.  The abilities to forecast demand and supply, prices and profit margins, raw material and operational costs, and customer flow and employee staffing needs could all help organizational decision makers choose the courses of action needed to best benefit their organizations.  Unfortunately, the ability to know the future is not something that human beings possess.

 The management science field has tools and techniques that can be used to solve many types of management and organizational problems—such as, inventory management, production and process control, product mixing and blending, transportation assignment and scheduling, maximization and minimization, and employee scheduling.  There are also forecasting techniques that decision makers can use to make educated guesses about their futures.  Linear regression is one of the most popular and well known of those techniques. 

Regression analysis is grounded in the concepts of correlation.  Correlation is a measure of relationship between variables.   Variables can be positively or negatively related or not related.  A positive relationship means that as one variable increases in value, its associated value does as well.  For example, a higher number of sales calls placed should result in higher number of sales made.  A negative relationship means that as one variable increases, the other variable decreases.  As the number of days missed from work increases for a worker, the performance for that person should decrease.  A zero correlation, or no correlation, indicates that two variables are not related—such as the number of soft drinks consumed at work and the number of maintenance calls made for copy machine repairs.

When a relationship exists between two variables, it is possible to predict one with the other.  The stronger the relationship between the variables, the better the predictive power.  For example, when fuel efficiency of a vehicle is known, it is easy to forecast fuel consumption when distance is known.  When a known relationship is not as strong, such as the effects of an advertising campaign on the sales of an item, the ability to predict the outcome is not as certain.  Regression analysis measures the relationships between variables and builds a mathematical model to best describe those relationships—which can be used to predict an outcome variable with one or more predictor variables.  With education and training on regression techniques, which can be run in Microsoft Excel, managers can forecast the future for their organizations.

<Back to Articles Page