Support Vector Regression
Definition of Support Vector Regression:
Support vector regression (SVR) is a machine learning algorithm that can be trained to learn the nonlinear relationship between input data and a target output variable without prior description of the underlying processes. SVR is based on support vector machine theory that classifies data by representing them in a multidimensional space.
This is the common definition for Support Vector Regression, other definitions can be discussed in the article
|
Short introduction
A support vector machine is a classifier that maps a subdomain of the input variables onto a subdomain (a class) of the output target variable(s). If the dependence of the output variable(s) on the input variables is nonlinear, it is not possible to define a linear classifier that separates input variables and corresponding output variables in distinct classes. By performing a so-called nonlinear kernel transformation, a low dimensional data space is converted into a high dimensional space where a linear hyper-plane can classify the data points (i.e. define distinct subdomains of input data and corresponding output data). Given the kernel function, the support vector machine does a systematic search to determine the hyperplane that best separates the training data into different input-output classes. Support vectors correspond to the data points that are near to the hyperplane and help in orienting it. Once the hyperplane is known, the position of a new input data point (e.g. from the test data) relative to the hyperplane determines the class to which it belongs. Support vector regression assumes that classes correspond to certain domains (value ranges) of the target variable. A number of kernel functions exist such as Polynomial Functions (mapping data onto a finite-dimensional space) or Radial Basis Functions (mapping data onto an infinite-dimensional space) that enable non-linear classification. Support Vector Regression further assumes that the training and test data are independent and preprocessed in order to follow identical distributions (e.g., subtraction of the mean value and division by the square root of the variance). Being a highly sophisticated and mathematically sound algorithm, Support Vector Regression is one of the most accurate machine learning algorithms.
For more detailed explanations see:
- StatQuest: Support Vector Machines Part 1: Main Ideas by Josh Starmer
- Wikipedia Support Vector Machine
Please note that others may also have edited the contents of this article.
|
