In this video we examine Decision Trees. These are perhaps one of the most simple, intuitive yet most powerful techniques in machine learning.
Hit '>Play' to learn more about how Decision Trees - one of the most open and 'White-Box' AI tools
Decision Trees are most often used as classifiers, to identify which group or class some input data should belong to. We are going to use them to identify which loan applicants are most likely to default. But they can also be used for regression tasks where we want to predict an output number from an array of input parameters.
Decision trees are an example of a supervised learning algorithm. We train them with some data that includes a target variable. Whether a borrower defaulted or not. The decision tree learns to ask a series of binary yes/no questions on the rest of the training data that allow it arrive at the correct classification.
While some AI techniques deliver impressive results, the way they achieve their performance is unclear. People are naturally suspicious of black-box AI. Neural Networks are a great example here. In contrast Decision Trees are white-box solutions. You can see each decision rule in plain English. Everyone can understand how the tree is classifying the data.