Second meeting: Introduction to Statistical Learning (ISL) study group
At this meeting we will be joined by two guest presenters:
- Filipe Santos (Senior Data Scientist with Liberty Seguros) will take us through an simple example of using Gradient Boosting Machines (GBMs) with insurance data. We are going to tune our model and look under the hood to be able to understand better the so called “black box model”.
In advance of this meeting Filipe Santos has provided a preview of his presentation on GBMs. This is available at https://github.com/saidataanalytics/ISL/tree/main/gbm_26may. The presentation is in ELS Training.nb.html - you may need to download the file to your computer and then open it from there.
As part of his presentation on GBMs (Chapter 8 of the textbook/video series), Filipe will be using cross-validation which is discussed in Chapter 5.
- David Robinson (RSA) will discuss a clustering and geographical visualisation example using census data.
The parts of the course that are most relevant are Chapter 8 and Chapter 10 (in particular, the K means clustering sections) in both the book and the video lecture series
- Videos available at https://www.dataschool.io/15-hours-of-expert-machine-learning-videos/
- Book available at https://www.statlearning.com/
The code from the two presentations on May 5 is also available on this site:
- Classification example: https://github.com/saidataanalytics/Classification_Worked_Example
- https://github.com/saidataanalytics/lassoexample
The Society has formed a study group of interested individuals to work through the material in the Introduction to Statistical Learning textbook and lecture series. This provides an introduction to machine learning from a statistical background and will be of interest to those who are new to the area.
More information is available at https://web.actuaries.ie/forum/r-and-python-study-groups/next-online-course-starting-introduction-statistical-learning, or by email to info@actuaries.ie (who will pass it on) for those who don’t have access to the forum or would prefer not to use it. You are also welcome to ask questions during the zoom meetings.
If you are not registered for this study group, but would like to join, please 'Make a reservation' below.
NB: As a security measure, there is a two-step registration process before you receive the link to join any webinar or meeting. The first step is to register your interest in attending a webinar or meeting by making a reservation for an event via the Society’s website. You will receive a link to then register via the Zoom platform. It is after this second registration step that you will receive the link to join the webinar or meeting.