Type I and Type II Errors: What's the Difference? - KDnuggets

Let's illustrate Type I and Type II errors using a binary classification machine learning spam filter. We will assume that we have a labelled dataset of N 315 emails, 244 of which are labelled as spam, and 71 are not-spam. Supposed that we've built a machine learning classification algorithm to learn from this data. Now we would like to evaluate the performance of the machine learning model. How good was the model in correctly detecting the spam vs not-spam emails? We will assume that whenever the model predicts an email to be a spam email, the email will be deleted and saved in the spam folder.