Boosted Tree Model
This is an ensemble decision tree model built using boosting technique on gas sensor array data. Adaboost algorithm was used to build the ensemble. Two variations of the model were built. In the first model Brieman coefficient was used to update the weights whereas the second model used Zhu coefficient. Results are presented below.
Boosted Tree Using Breiman Coefficient
Misclassification Rate - 0.5% (on test data)
Confusion Matrix
Boosted Tree Error As a Function of Number of Iterations
Boosted Tree Using Zhu Coefficient
Misclassification Rate - 0.5% (on test data)
Confusion Matrix
Boosted Tree Error As a Function of Number of Iterations
ROC Curve
Area Under Curve
0.9995898