cowboyuniverse / cancerprediction

Cancer Prediction Using Random Forest and Decision Trees

cancerPrediction

Cancer Prediction Using Decision Trees and Random Forest

Bagging Accuracy 0.822222222222

AdaBoost: 0.933333333333

AdaBoost accuracy with cross validation: 0.929464285714

Accuracy Random Forest: 0.955555555556

RandomForest accuracy with cross validation: 0.929464285714

• Making a bootstarp sample of the original “Training” Dataset (build in part(b)) with the size of bootstarp_size = 0.8*(Size of the original dataset). You can use the following command to generate a random bootstrap dataset (“i" is the variable of the loop, so the random_state changes in each iteration): resample(X_train, n_samples = bootstarp_size , random_state=i , replace = True)
• Define and train a new base decision tree classifier on this dataset in each iteration: Base_DecisionTree = DecisionTreeClassifier(random_state=2).
• Test “this base classifier” on the original “Testing” Dataset build in part(b), and save the prediction results for all testing samples.
• Perform Voting to make the final decision on each data sample based on the votes of all 19 classifiers. Finally, calculate and report the accuracy of your Bagging method.

author: yosep kim

github.com/cowboyuniverse