Machine Learning: final assessment
This covers the whole course. Take it after the last module.
Final quiz
Practical assessment: An end-to-end classifier
Without hints or libraries, write a program that trains and evaluates a classifier on a small table of data you create, from raw rows to a final test score.
- Missing values are filled and features scaled with statistics from the training data only
- At least two models are compared on a validation set, one of them a majority-class baseline
- The chosen model's settings (such as k, λ or the threshold) are tuned on validation data
- The final model is evaluated once on the test set, with a confusion matrix, precision and recall
- Accuracy is reported separately for at least two groups in the data
Final project
A spam classifier, compared three ways: Build a spam filter for short messages with three different methods and decide which one you would deploy, with evidence.
How it is evaluated
- Correctness: each method and metric gives results you can check by hand on tiny inputs
- Honesty: no information from the test set is used before the final evaluation
- Understanding: you can explain every choice and every number in the report
- Clarity: preparation, models and evaluation are separate, readable functions
To complete this course
- Score at least 70% on the final quiz
- Solve the beginner and intermediate exercises in every lesson
- Finish one project and check it against its evaluation criteria