Workshop details

• Machine learning is a powerful tool for analyzing data and make striking predictions when good quality annotated datasets are available. With the plethora of health datasets made available by health providers and research projects world wide, it is indispensable for the modern data scientist to master in such techniques. We will use R in this workshop to demonstrate how such analysis can be conducted using Caret, a dedicated ML library for ML solutions in R.

• It is recommended that attendees not familiar with machine learning concepts attend the preparatory workshop Introduction to Machine Learning for Health Data Science a priori.

Learning objectives

• Get familiar with RStudio for the implementation of ML code;

• Load and explore tabular datasets for ML;

• Explore the distribution and correlations among the dataset's variables;

• Load and use Caret package for train and test ML models;

• Evaluate and compare the prediction results of various ML methods.

Syllabus

• Introduction to the Diabetes dataset;

• Preparing the RStudio environment;

• Loading and exploring the dataset;

• Distribution plots;

• Scatter matrix plots;

• Features correlation analysis;

• Introduction to Caret;

• Dataset split;

• Model tuning;

• Learning curves;

• Logistic Regression;

• Decision Trees;

• Support Vector Machines;

• Classification report;

• Confusion matrix;

• ROC Curves;

• Methods comparison.

Format

• Zoom live webinar: 1h live webinar with pre-recorded presentation split in two sessions of 30min each with a break of 5min between them. Attendees are welcome to join asynchronous live group discussion (Q&A) with the speaker on Discord during and after the scheduled session.

• General discussion session: 1h Q&A recurrent pre-dated session booked for attendees of multiple slots of this workshop.

• Recorded presentation: Attendees registering using a Google e-mail account, will have private access to the recorded webinars.

• Certificate: all attendees checking-in and out of both 30min sessions will be entitled to receive a certificate of participation (PDF/PNG) that can be shared on your social media with credentials (QR-code) that can be verified on our website, with an option (extra fee) to register as a NFT asset with an unique address on the Ethereum blockchain.