Support Vector Machine Classification
Support Vector Machine (SVM) is an advanced supervised modeling technique for classifying both linear and nonlinear data. SVM Classification clusters data into the most distant and distinct groups as possible.
SVM has become a more recent default approach to classification problems because it is well suited to very high-dimensional spaces and extremely large datasets. The main idea behind Support Vector Machines is to maximize the distance between the classification datasets and a subset of training data points called support vectors.
- Support vectors are the closest points from every class to the decision boundary.
- The TIBCO Data Science - Team Studio SVM Classification operator handles the classification of both linear and nonlinear data.
- Support Vector Machines, using common Kernel Method transformations, are able to project data that is noisy or not linearly separable, into a higher dimensional linearly separable space.
The TIBCO Data Science - Team Studio SVM Classification operator allows for multi-class classification (as opposed to only binary classification) on both categorical and continuous numerical data. Unlike with the CART, Decision Tree and Random Forest classification operators, there is no tree-like branching structure created with SVMs; rather, the defining details of the Support Vectors that best separates the data are provided.
SVM Classification modeling is often used for complex artificial intelligence applications, such as object detection (computer vision), speech recognition (NLP), handwritten digit recognition, text classification, and bioinformatics (for example, gene classification).
Advantages of Support Vector Machine modeling include:
- Ability to handle high dimensionality of large datasets.
- Ability to model complex nonlinear decision patterns.
- Known for being extremely accurate in many cases.
- Less prone to over-fitting than other methods, like a simple Decision Tree or Naive Bayesian classifiers.
- Provides a compact description of learned model by only using the Support Vector data points for each class.
To use SVM in your workflow, see its operator help at SVM Classification .