Machine Learning: Understanding Decision Tree Learning

As the data that is fed becomes larger, the decision tree tends to become longer. In such cases, noise and corrupt/incorrect data can have a detrimental impact on the decision tree. This results in the decision tree overfitting the dataset, that is decision tree performs satisfactory for the training data, but fails to produce an appropriate approximation of the target concept when it encounters actual data. Overfitting can also occur when insufficent data is provided to build the decision tree (like perhaps, our previous with only 6 rows.)