Is the Model Actually Good? Baselines
A score means little without something to compare with. A baseline is the simplest possible predictor, such as always answering the most common class. If 95% of emails are not spam, a "model" that always says "not spam" is 95% accurate and completely useless. A real model must clearly beat its baseline, and on the cases that matter.
Examplepython
Quick check
A model is 95% accurate, but 95% of the data is one class. What should you check?
Your turn: write the code
Print the baseline's accuracy on a dataset with 90 negatives and 10 positives.
Finish the quick check (and any code exercise) to complete this lesson, or mark it done yourself.
Tip: use the left and right arrow keys to move between lessons.