Detecting outliers using iForest
iForest has similarities with another popular algorithm known as Random Forests. Random Forests is a tree-based supervised learning algorithm. In supervised learning, you have existing labels (classification) or values (regression) representing the target variable. This is how the algorithm learns (it is supervised).
The name forest stems from the underlying mechanism of how the algorithm works. For example, in classification, the algorithm randomly samples the data to build multiple weak classifiers (smaller decision trees) that collectively make a prediction. In the end, you get a forest of smaller trees (models). This technique outperforms a single complex classifier that may overfit the data. Ensemble learning is the concept of multiple weak learners collaborating to produce an optimal solution.
iForest, also an ensemble learning method, is the unsupervised learning approach to Random Forests. The iForest algorithm isolates anomalies by randomly partitioning...
