****************Use R******************************************************

Use dataset credit-g

https://archive.ics.uci.edu/ml/datasets/Statlog+(German+Credit+Data)

Use R and the caret package to tune a good classification model to predict credit risk and to evaluate the quality of the model. Try decision trees (rpart in the caret train function).

1. Experiment with the parameters of the decision tree algorithm and determine a good setting for the parameter(s). Can you determine the best parameter value for decision tree learning when applied to this dataset?

2. Experiment with different methods for estimating errors. Specifically try using a) an independent test dataset, b) cross-validation and c) bootstrap. For cross-validation try different numbers for the number of folds. For bootstrap, try both the simple bootstrap (method boot in the train function) and 0.632-bootstrap (method boot632 in the train function). Can you determine the best method for estimating the error?

3. Write a summary. Explain why you believe your parameter values are good. Also explain which method for estimating the error is best and why.