Machine Learning on a Cancer Dataset - Part 34

This is the final tutorial in which we discuss uncertainty estimation for machine learning in scikit-learn.

In the previous two videos we delved into uncertainty estimation by looking at the 'decision_function' and 'predict_proba' methods for binary classification. We used a support vector machine and we looked into uncertainty estimation for binary classification: categorizing tumor samples as malignant or benign.

What happens if we work with a dataset where data is labeled in more than two classes, so a non-binary situation. That's what we're getting into in this video tutorial.

First of all, we're using a different classifier - a GradientBoostingClassifier. Since we cannot work with the breast cancer dataset (because of its binary labels), we're using the 'iris' dataset, which, similar to the cancer dataset, is preloaded/preprocessed in scikit-learn. The 'iris' dataset categorizes flowers into: setosa, virginica, and versicolor.

So, let's see how uncertainty estimation looks like for this type of multiclass classification. Please watch the video below for the complete walkthrough.

Previous videos in this series:

1. Machine Learning on a Cancer Dataset - Part 30
2. Machine Learning on a Cancer Dataset - Part 31
3. Machine Learning on a Cancer Dataset - Part 32
4. Machine Learning on a Cancer Dataset - Part 33

To stay in touch with me, follow Hive account@cristi

Cristi Vlad, Self-Experimenter and Author