Machine Learning in Python with scikit-learn (3 sessions)

Tutor: Dr Irina Chelysheva. Contacts: Oxford Uni profile, Twitter

In this short series we will get familiar with the most common Machine Learning algorithms and apply them in Python, using scikit-learn and beyond. We will perform end-to-end ML projects for various data types and research questions. We will select the best ML method by evaluating their performance, use feature selection approaches, apply cross-validation and make the actual predictions. For those previously participating in the first two sessions – new(!) third session will be covering unsupervised ML methods and their applications.

Topics to be covered

• Overview of ML methods and algorithms, unsupervised vs supervised learning

• Evaluation of the performance of the algorithms and the choice of the best one

• Application of ML to different problems – classification vs regression, different types of data

• Hands-on end-to-end ML analysis in Python with sklearn

• Feature selection methods and their application

• Metrics and evaluation of the performance – using cross-validation and making further predictions

Learning Objectives:

• perform end-to-end machine learning analysis of the dataset using Python;

• select the best machine learning algorithm for particular dataset;

• apply ML methods to both regression and classification problems;

• perform the feature selection;

• make the predictions and evaluate the performance of the ML algorithms

Software required

Python3 - https://www.python.org

Python SciPy libraries:

• scipy

• numpy

• matplotlib

• pandas

• sklearn (shorten from scikit-learn) - https://scikit-learn.org

Any suitable IDE (I use Spyder - https://www.spyder-ide.org)