|
Abstract:
|
Machine learning is becoming appealing in statistics for a number of reasons. First, the statistical properties of some learning machines are better understood. Second, traditional statistical approaches such as high throughput molecular biology technologies often fail with Big Data. Third, several machines have been extended to operate beyond the standard classification problem for dichotomous endpoints. Many statisticians are, however, not familiar with recently developed machine learning approaches such as gradient boosting, random forests, or support vector machines and their extensions. This course therefore aims to provide an introduction to some of the most important machine learning approaches currently used. We show that all problems from generalized linear models, and even survival endpoints, can be tackled with machine learning. The focus of the theoretical sessions is the nontechnical, but intuitive, explanation of the algorithms (instructor: Andreas Ziegler), and the focus of the hands-on laptop sessions is to see the machines operating using R (instructors: Andreas Ziegler and Marvin Wright). The combination of simple descriptions in a language familiar to statisticians together with the use of standard statistical software should help to demystify machine learning.
|
