SAS HPSPLIT: A Powerful Machine Learning Tool
This talk will explain a powerful, and FREE with SAS Stat., Machine Learning technique: HP Split. HPsplit can be coded to run a decision tree (to predict Categorical variables like buy/not-buy, etc.) or as a Regression Tree ( to predict a continuous variable like dollars spent or probability of an event). It is simple to use, fast to run and should be in every analyst's tool kit. Using this requires very little math and the power of this technique should not discourage anyone from learning how to use it.
The talk will cover the following topics: The branching logic of trees, the history of trees, using cross validation to reduce over-fitting, measures of model "goodness-of-fit and variable importance.
The talk will cover the following topics: The branching logic of trees, the history of trees, using cross validation to reduce over-fitting, measures of model "goodness-of-fit and variable importance.
An Animated Introduction to Git and GitHub
Git is used for version control of programs and is both commonly used and complicated. The true Git process involves storing "program changes" in a series of linked compressed files. Git is often explained by means of three high level abstractions: "Working area", "staging area" and "commit area". These high level abstractions are good when explaining basic Git. However, these abstractions only approximate what is really happening in Git and the abstractions break down when one tries to understand more complex Git actions. Additionally, since the internals of Git change as the user issues commands, verbal explanations of Git, or static representations of the Git system can be ineffective. This cartoon-like animation of the underlying Git processes is an easy and quick way to understand what Git really does and opens the way to viewing and understanding more advanced Git actions.
|
About the Presenter
Russ Lavery is a frequent and multiple award-winning presenter at SAS and Stat conferences all over the US, in Europe and Asia. He has been a technical reviewer on five books on SAS and statistical topics, including two books on Machine Learning. He has over 25 years of experience using SAS and is still studying. Russ is a contractor and lives outside Philadelphia, PA, where he occasionally teaches as an adjunct in the Drexel University analytic program and dances frequently.
|
