Random ForestConclusion
Random forests remain one of the most popular methods in classical machine learning, and now you know why they work so well.
- Strength in numbers. As Condorcet's jury theorem says, a majority vote of many models that are each better than a coin toss can be right far more often than any one of them.
- Diversity is the key. Bagging and random feature choice force the trees to be different, so that their mistakes are different.
- Variance cancels out. One deep tree overfits and changes with every small change in the data; the average of many such trees is steady.
In short: deep trees are clever but wobbly, and voting keeps the cleverness while calming the wobble.
Speed Round: Bags and Votes
Get five right to move on.