Algorithms for Machine Learning

Anomaly Detection with Principal Component Analysis

Use Case

Before you explore the details of anomaly detection with principal component analysis (PCA), you should first know when to use this algorithm. A typical use case would be the following:

For more information, see Anomaly Detection with Principal Component Analysis (PCA).

Distance-Based Failure Analysis Using Earth Mover’s Distance

Use Case

Before you explore the details of distance-based failure analysis using earth mover’s distance (EMD) as known in computer science,(also known as the Wasserstein metric in mathematics), you should first know when to use this algorithm. A typical use case would be the following:

For more information, see Distance-Based Failure Analysis Using Earth Mover’s Distance (EMD).

Remaining Useful Life Prediction Using Weibull

Use Case

The Weibull algorithm can be used to calculate the expected remaining useful life (RUL) of an asset, and to calculate the probability of failure of an asset. A typical use case would be the following:

For more information, see Remaining Useful Life Prediction Using Weibull (RUL).

Anomaly Detection Using Multivariate Autoregression

Use Case

The algorithm for multivariate autoregression (MAR) can detect dependencies between different kinds of sensors even if the influence of one sensor by another one is delayed over time.

MAR produces good results if abnormal behavior is detected for systems with various different kinds of sensors that (partly) depend on and influence each other.

For more information, see Anomaly Detection Using Multivariate Autoregression (MAR).

Failure Prediction Using Tree Ensemble Classifier

Use Case

The TEC algorithm can be used whenever the following applies:
  • The goal is to predict failure of a system under investigation.
  • Historical sensor data or feature records for the system or a similarly behaving system are available.
  • Feature records include labels for each record whether or not the record corresponds to a failing system.

For more information, see Failure Prediction Using Tree Ensemble Classifier (TEC).