Testing ML

---

• Create output that is easy to compare by computer (so numerical results are preferable over a graph)
• Fix randomizations to make the results repeatable

• Establish thresholds for results using different datasets (and also using different models)