Ml-study 20: data modeling

Here are our plans for the coming weekend.
Please write to me about which session(s) you are planning to attend.
Note that some knowledge will be assumed – see below.

Follow the #ml-study Zulip stream for details and discussion.

ml-study 20: data modeling

This weekend at #ml-study we will practice data modeling using R4DS Section 24: Building models.

Sessions

We may organize sessions for the following time slots.

The three sessions will be independent.

Agenda

In each session, we will practice some parts of the R4DS section in Clojure.

Assumed knowledge

  • We will assume familiarity with scicloj.ml. You can either listen to the ml-study recording 18.2.
  • We will assume everybody has read R4DS Section 24: Building models. There is no need to understand all the technical details about R. The important parts are the modeling ideas and the exploration of the datasets.
  • It is recommended to review or read about the basic notions of supervised learning, such as training/validation/testing, overfitting/underfitting, hyperparameter tuning, etc.

Method

One person will be sharing the screen and exploring in their machine. We will help that person and think together as a group. If you wish to be that person, please reach out to me beforehand.

Practice

After the sessions, everybody is invited to pick some data problems or existing machine learning tutorials and practice further.

Recording

As usual, we will record the sessions but share the recording only here, for the friends who miss the session.