The following considerations when using the Decision Tree model are worth noting:
- If an SQL view is used for input, and if that view references a large table, the repetitive nature of the Decision Tree algorithm causes multiple readings of that large table.
- If the view is persisted to a table of limited size (for example, 50 or 100 rows), then the runtime becomes much faster.
- Using the max_depth parameter for data that is not well-structured or random helps shorten the runtime.