The materials on this page are aimed largely at novice learners in machine learning without a strong quantitative background. The emphasis of the materials is on understanding the process and related issues and problems that might arise, rather than mathematical rigor.
It can be taught entirely without mathematical derivations (e.g. at a secondary-school level), but some derivations exist (primarily in the logistic regression lesson) that may be appropriate for slightly more advanced students. Required mathematical concepts in the latter case include functions and parameters, a basic understanding of the process of optimization, exponentials and logarithms, sums and sets, square roots, geometric and harmonic means.
Time estimates for taught components and exercises assume a college-level audience, without in-depth mathematical derivations or computational exercises. Instructors of more junior audiences are advised to remove material and double the time for exercises and discussions. Exercises that involve programming tend to be time-consuming, unless the audience is very computationally proficient. Allow at least an hour for each exercise involving programming.
At the high-school level, the primary objectives for teaching this particular class are (1) to demystify machine learning and make the subject matter approachable to students without a strong computer science background, and (2) educate students with the goal that they may subsequently understand the effects and limitations of machine learning systems that affect their daily lives in numerous ways. In the latter context, it may be beneficial to devote more time, and an even stronger emphasis, on the ethics lesson in Episode 3, and devote extra discussion time to the topic here. At the high-school level, the two algorithms can be taught visually, without going too much into the mathematical details.
At the college level or above, we expect that learners are more likely to want to implement their own machine learning models either for fun or for some practical purpose. Here, instructors may wish to place an emphasis on the limitations of machine learning and on practical advice for building ethical machine learning systems. Teaching the mathematical concepts may be beneficial, but instructions should certainly include a discussion of the limitations of the algorithm, how to assess these limitations and compare them to the training data.