Decision Tree Classification: An Important First Step (Poster 1)

• In Event: AERA Poster Session Saturday 3:20 p.m.
  In Poster Session: SIG-Educational Statistician Poster Session

Abstract

Decision Tree uses variable information gain values to decide which independent variable to use for the first node in the decision tree. This first step is automatically selected by the Decision Tree algorithm. If an effective set of independent variables are not chosen for the Decision Tree analysis, then the fit function (classification accuracy) would not yield the best accurate results. This useful “fit” function is displayed in the confusion matrix. Decision Tree modeling could benefit from an important first step. Prior to Decision Tree analysis, a Principal Component factor analysis with a single factor could be run to indicate which variables have the highest variance. An example is presented.

Authors

• Randall E. Schumacker, University of Alabama

• Todd Sherron, Texas State University