You are on page 1of 5

RANDOM FOREST

CLASSIFIER
ALEXANDRU-FLORIN GAVRIL
INTRODUCTION

Random forests are an ensemble learning method for classification


Constructs a large collection of single decorrelated decision trees and
uses a voting system on the output
Created by Tim Kam Ho, then extended by Leo Breiman and Adele Cutler
which combined the initial random selection features idea presented by
Ho, combined with Breimans bagging idea
Rationale

The combination of learning models increases the classification accuracy


The average noisy and unbiased models are used to create a model with
low variance
How it works?


Let , , ... , be the training set, with = [, , , ] a sample with k features,
each having a response (class)
Split the training set into M subsets of training samples
Construct decision trees with each of the M subsets by choosing a
random attribute every time.

The decision trees will each classify a sample and then will provide a
response which will enter a voting process and choose the
corresponding class.
How it works?

You might also like