Random forest

How can we make a forest of decision trees smarter and more accurate?

Imagine you're trying to predict if a new coffee shop will be popular in your neighborhood. You could look at many factors like location, price, and menu variety, but it's hard to know which ones matter most.

Think of Random Forests as a group of coffee shop experts, each focusing on different factors like location or price. They all vote on whether the shop will be popular, and you take the majority vote as your prediction. This way, you're not relying on just one expert's opinion, which might miss important details.

Example

Let's say 10 experts (trees) predict the success of a coffee shop. 7 say it's likely to succeed because of its great location, 2 think it's too expensive, and 1 is unsure. The majority vote is success, so you predict the shop will do well.

Remember this

Random Forests combine many experts' opinions to make a better prediction, reducing the chance of getting it wrong.

Text adapted from Wikipedia, licensed under CC BY-SA 4.0.

Related concepts

to use random forests: when you want a strong baseline with minimal hyperparameter tuning

Why not always use the best single tree for predictions?

Boosting (machine learning)

Boosting reduces bias in ML models

to use AUC-ROC: comparing classifiers across all thresholds

Why can't we always trust a yes-or-no answer?

score matching does: learns the gradient of the log-density without normalizing

Ever wonder how we can compare apples and oranges fairly in studies?

Overfitting

Why does fitting too much detail to data ruin predictions?

Convolutional neural network

Can a neural network learn too well?

Swipe through more Machine Learning concepts

Open Pocket Polymath