Predicting Gas Mileage using the T-Method

The T-method is a technique developed by Genichi Taguchi to calculate an overall prediction based on the signal-to-noise ratio without the use of Gram-Schmidt orthogonalization. The Taguchi methods, also known as Robust Design principles, are used for optimization through planning, conducting, and evaluating the results of experiments to determine the best levels of control factors. The primary goal of Robust Design is to minimize variance in the presence of noise factors to achieve a robust process. This paper outlines the T-method steps using a forecasting case study to calculate miles per gallon for vehicles with a univariate response to illustrate the technique. The methodology used to forecast in the case study is outlined. In addition, a basic comparison with the Mahalanobis-Taguchi System is provided.