Decision Stumps

Question

I'd like to implement a java application using AdaBoost wich classifies if an elephant is African or Asian Elephant. My Elephant class has fields:

int size;
int weight;
double sampleWeight;
ElephantType type; // (which can be Asian or African).

I'm a newbie with AdaBoost and I've learned that good weak classifiers are decision stumps. I'd like to know if I am suppoused to create only 2 decision stumps(1 for size and 1 for weight) or should I make more decision stumps(few different for size and few different for weight)? How does exactly creation of classifiers look like?

Answer 1

You will create many decision stumps. The point of boosting is that each time you create a new classifier (in your case, decision stumps), you will increase the weights (importance) of all instances that the classifier misclassifies so that that the next classifier focuses more on the misclassified instances. You can randomly select which your two features you use in your stump for each step. Once you have create your entire set of classifiers (stumps), they use a majority vote to classify new instances.