Isolation Forest#
An ensemble learner comprised of Isolation Trees that are each trained on a unique subset of the training set. Isolation Trees are a type of randomized decision tree that assign anomaly scores based on the depth a sample reaches when traversing the tree from root to leaf node. Anomalies are isolated into their own nodes earliest during tree traversal and therefore receive the highest isolation scores. The Isolation Forest works by averaging the anomaly scores for an unknown sample across a user-specified number of trees.
Interfaces: Estimator, Learner, Ranking, Persistable
Data Type Compatibility: Categorical, Continuous
Parameters#
| # | Param | Default | Type | Description |
|---|---|---|---|---|
| 1 | estimators | 100 | int | The number of isolation trees to train in the ensemble. |
| 2 | ratio | float | The ratio of samples to train each estimator with. If null then the subsample size is 256. |
|
| 3 | contamination | Auto | float | The proportion of outliers that are presumed to be present in the training set. |
Additional Methods#
This estimator does not have any additional methods.
Example#
use Rubix\ML\AnomalyDetectors\IsolationForest;
$estimator = new IsolationForest(100, 0.2, 0.03);
$estimator = new IsolationForest(100); // Default sample size and threshold
References#
- F. T. Liu et al. (2008). Isolation Forest.
- F. T. Liu et al. (2011). Isolation-based Anomaly Detection.
- M. Garchery et al. (2018). On the influence of categorical features in ranking anomalies using mixed data.