Assessment of Fuel Consumption Using Factorial Experiments, Regression Trees and MARS

Abstract

The purpose of this paper is to determine the effects that are significant on the fuel consumption of F-4 aircrafts by using three different statistical methods and compare the results of those different methods. The response variable of every possible combination of factors and interactions are analyzed to provide information about main effect(s) and interaction effect(s). MARS is applied in order to find the relationship and a tree model in order to summarize the relationship of main effects and the interactions on a plot. Also 2(5) factorial experiment with 4 replications is performed to determine the significant main effects and interactions. A comparison of performance of the methods is made in order to investigate their applicability.