[TriZPUG] fw: Intro to scikit-learn (Machine Learning in Python) with Tim Hopper

[Tom Roche]

Raleigh-Durham-Chapel Hill R Users Group Mon, 13 Jan 2014 11:29:27 -0500
> Intro to scikit-learn (Machine Learning in Python) with Tim Hopper
> Raleigh-Durham-Chapel Hill R Users Group
> Thursday, January 16, 2014 6:30 PM
> Cameron Village Regional Library
> 1930 Clark Avenue
> Raleigh, NC 27605

> There has been a lively debate recently on whether Python is poised to supplant R as the preferred tool for data scientists.  David Smith [an R vendor] sums it up well at

> http://blog.revolutionanalytics.com/2013/12/r-and-python.html

> This Thursday we have an opportunity to learn more about how Python can be used for data work.  Tim Hopper will present Scikit-learn, a Python package providing an implementation of many machine learning algorithms that R can also do (e.g. SVM, kNN, linear models, HMM, k-means, spectral clustering).

> Description from Tim: "The benefits of Scikit-learn goes well beyond carefully implemented learning algorithms. Being built in Python, it allows easy integration with countless other Python modules for tasks such as plotting, data munging, and application development. Its consistent API across algorithms allows for rapid experimentation with multiple learning methods. Also, Scikit-learn is well documented and provides lots of examples.  Instead of discussing particular machine learning algorithms provided by the package, I will focus on Scikit-learn and Python as a toolkit for solving data problems from start to finish. I will emphasize the Pipeline tool which allows the user to chain together all the steps of a machine learning pipeline including preprocessing, dimensionality reduction, feature selection, and model fitting."