Revolutions

Data Science Education gets personal

by Joseph B. Rickert

It is difficult to imagine that there is anyone on the planet with an internet connection and a desire to learn something new who has not at least looked into taking a massive open online course (MOOC). Last Fall, in an 11/4/12 article, the New York Time declared the Year of the MOOC and quoted one of Coursera’s founders, Andrew Ng, of Stanford Machine Learning fame, as boasting that with over 1.7 million students “Coursera was growing faster than Facebook”. This year with both Udacity and Harvard and MIT backed edX offering interesting and challenging courses the growth of MOOC enrolment must be astounding indeed. I think this is all good and I have written about it before .

Being the perpetual student, I have looked over the syllabi of quite a few courses that seem to be pretty exciting, but I am struggling to carve out the time to devote to them.  MOOC courses are “free”, but for a working professional they not without opportunity costs. In fact, it is not clear that a course designed to add value to 100,000 plus people at one time is the most effective way for professionals working in or around data science to make progress.

Tony Ojeda of the R-users-DC saw the interest professionals have in learning more about data science, and realized that even with the prevalence of MOOC courses there are educational needs that are not being met. In a recent email Tony wrote:

I started attending data meetups like R Users DC and Data Science DC about a year and a half ago … and these events really made me realize that there are a bunch of people out there who, like me, wanted to learn more about data science, machine learning, and analytics.  However, I noticed that there isn't really a set path for people that want to get started learning data science.  It all depends on your current context and learning how to apply the methods and tools to what you're currently doing.

Tony also noticed that many of the people attending the meetups were very knowledgeable about statistics, programming or other aspects of data science. And, as he describes it:

I'm sort of an efficiency freak, so I automatically thought, “It would be awesome if I could just find someone who knew more than me about whatever I'm trying to do so that I could get past these hurdles and get my work done." 

The result of Tony’s musings was that on January 9^th of this year he announced www.sagebourse.com, a bare bones site for matching students with data science tutors in the greater DC area (including Maryland and Virginia). Students accessing the sight can express and interest in being tutored in either Data Science or Programming. Currently, Data Science topics are limited to Statistics, R, Machine Learning, Matlab/Octave or Excel.  The programming languages listed are: Python, Ruby, Java, Javascript, C, C++, Perl and PHP. In addition to these topics, sagebourse.com is also looking for tutors in SAS, SPSS, Stata, Hadoop, Access and Data Visualization.

Tutors who register with the sagebourse.com “bid” on the gig by providing the hourly rate at which they are willing to teach a subject. When students register with the site they fill out a form indicating the subject they want to be tutored in, provide contact information and indicate where they would want the tutoring to take place. Students then receive an email with bids from several tutors. Presently, the process of matching tutors with students takes a few days depending on the responsiveness of tutors and students.

These are still very early days for sagebourse.com. About 20 tutors have registered with the site and according to Tony they have already had a “handful” of successful tutoring sessions. Statistics and R are by far the most popular areas of expertise professed by the tutors who have registered, and “In terms of demand from students, it has been R, Statistics, and Machine Learning in that order”.  

I am excited about what sagebourse.com is trying to do, and I hope that it can make a big, positive contribution to the R community.

The Education Revolution: Coursera's co-founder speaks

Joseph Rickert reports from last weekend's ACM Data Mining Camp in San Jose.

Andre Ng, co-founder of Coursera, Director of the Stanford Artificial Intelligence Lab and Associate Professor with the University’s Computer Science department gave the keynote address at the ACM’s Data Mining Camp held at the PayPal/eBay town hall this past Saturday. Andrew is on a mission. He is out to change the world by bringing relevant, high-quality education to anyone with an internet connection. Coursera has relationships with 33 universities and is currently offering 198 college level courses on subjects as diverse as Modern and Contemporary American Poetry, R and Machine Learning.

Although quality education via distance learning is hardly new, and MIT has been freely offering high-quality, college courses through its internet-based, OpenCourseware platform for over ten years, Coursera has broken new ground by offering an interactive learning experience on a massive scale. Recently over 300,000 students enrolled in the machine learning course. What makes this possible is the application of software development methodology, hard core programming and machine learning techniques to the challenges of developing assignments, grading homework exercises and building online communities. Tests and homework exercises go through unit testing and automatic proof checking, complex regular expressions evaluate free-form text, and every click from every student is captured and analyzed in much the same way that internet companies such as google and eBay capture information about their website visitors. Coursera is already analyzing some of this data to improve there product. For example, while reviewing answers to a machine learning assignment, Andrew noticed that 2,000 users submitted identical wrong answers to a programming assignment. A k-means clustering analysis revealed that the student errors originated with switching two lines of code in a particular algorithm. This information was used to improve the underlying lecture associated with the assignment.

The large numbers of students have had a positive effect on building communities. Andrew pointed out that it is easier to build a community with 100,000 students than with 100 students. With large numbers there is a high probability that any time of the day or night there will be someone else awake and thinking about the same issues you are. During the most recent machine learning class the average time to get a response to a question posted on the bulletin board was 22 minutes.  Andrew also noted that several study groups have spontaneously organized.

The large numbers of students also make it possible for Coursera to leverage crowd sourcing techniques. Using guidelines provided by course instructors, experiments with students grading each other have turned out well.

Andrew took several questions from the audience, some directly challenging the model of free, open source education. When asked if Coursera was competing with universities, Andrew replied that course content is only part of the value a student gets from attending classes at Stanford: students get more value from high quality classroom interaction and the attention of engaged faculty.

Andrew’s talk was inspiring. The classroom / lecture model for university education has not changed much from the time of Peter Abelard and Coursera could be the game changer. But even more importantly, the Coursera vision could drive a new enlightenment. Andrew and Coursera are taking the high ground: “A high quality education is a fundamental human right”.

A sociologist converts from Stata to R

Ph.D candidate in sociology Ethan Fosse just switched from Stata to doing 100% of his analysis with R. His reasons?

If you want to do Bayesian analysis or graph modeled coefficients (or work with complex data structures more generally), then R is much easier than Stata due to the object-oriented programming environment. It's unbelievably liberating to be able to save vectors, matrices, data frames, and so on from multiple data sources and manipulations in the same conceptual space. Additionally, R has fantastic graphics capabilities (3-D plots, rotating hyperplanes, social network graphs, and so on), offers excellent tools for analyzing and display so-called big data (for example, check out the "tabplot" command from Google), and is (frankly) a fun, intuitive programming language. If you need additional reasons to be an R convert, keep in mind that R is completely free, open-source, and extensible, with over 5,300 statistical packages (as of April 2012).

Read more about Ethan's conversion at the link below.

Culture Statistics and Society: I'm Converted to R Full-Time

Big Data statistics in the search for a cure for MS

Multiple Sclerosis (MS) is a debilitating and complex disease with an unknown cause — and for which there is currently no cure. The SUNY Buffalo is home to one of the leading multiple sclerosis (MS) research centers in the world, and as reported in Healthcare IT News, the research team is using IBM Netezza and Revolution R Enterprise to accelerate its study of more than 2,000 genetic and environmental factors that may contribute to MS symptoms:

Shawn Dolley, vice president and GM of global healthcare and life science at IBM Big Data (he worked for Marlborough, Mass.-based Netezza before IBM's acquisition of that company in 2010), remembers when Harvard Medical School researchers told him not long ago that, when it came to solving problems with big data analytics, "We have four questions we can ask per month." Why four? "Well, there's four weekends per month," he recalls them saying. "That's how long it takes us to run our jobs."

Now, using an IBM Netezza analytics appliance, in conjunction with software from Revolution Analytics, researchers can analyze disparate data in a matter of minutes instead of days, regardless of what type or size it is, say IBM officials. The technology automatically consumes and analyzes the data, and makes the results available for further analysis, leaving researchers more time to analyze trends. 

As described in this case study, SUNY Buffalo researchers were able to accelerate their research by reducing computation times more than 100-fold in the search for significant interactions between thousands of genetic and environmental factors. The R language itself also accelerates the research process, by making it easier to include new kinds of variables such as discrete dependent, Poisson dependent or continuous normally-distributed variables by simply adding a few lines of code. (Using another language would have required a re-writte of the entire algorithm.)

From an analytics perspective, the researchers were able to write a version of their software tools in in Revolution R Enterprise for IBM Netezza, so that all their reporting as well as our analysis is all consolidated in one place. This prevents the need to move very large amounts of data in and out of Netezza, which would cause further delays. They were also able to use a wider variety of data sets. 

The SUNY researchers were able to: 

• Use the new algorithms and add multiple variables that before, were nearly impossible to achieve. 
• Reduce the time required to conduct analysis from 27.2 hours without Netezza to 11.7 minutes with it. 
• Carry out their research with little to no database administration (Unlike other HPC platforms or databases available, Netezza was designed to require a minimum amount of maintenance)
• Publish multiple articles in scientific journals, with more in process.
• Proceed with studies based on ‘vector phenotypes’—a more complex variable that will further push the Netezza platform.

You can learn more about the research by the SUNY Buffalo team in this IBM Netezza blog post, and in the IBM press release at the link below.

IBM Press Releases: Leading Global Multiple Sclerosis Research Center Taps IBM Analytics to Improve Patient Care

R's continued growth in academia

Bob Muenchen has recently updated his report on the popularity of statistical software. With the updated analysis, we see that the R community remains as strong as ever: the number of contributed R packages continues its exponential growth rate, R maintains its dominance in online discussion, and has 20x the content of other statistics packages on social programming sites like CrossValidated and StackOverflow.

One particularly interesting metric is the number of times software packages are cited in scholarly articles, as a measure of the rate at which such packages are used in academia. Google Scholar makes it possible to quantify such measures, and from the chart below we can see that SAS and SPSS have been on a steep decline since around 2005 or so:

Amongst the software packages showing growth in academia (i.e. all those except SAS and SPSS), R is has the largest "market share" of citations, and continues to grow rapidly:

If you're interested in the details of how market share was calculated, or how to create a "market share" plot like this using the ggplot2 package, the librestats blog has everything you need at the link below.

librestats: Statistical Software Popularity on Google Scholar

Students: Help improve R with the Google Summer of Code

The Google Summer of Code project provides funding for students to write and contribute code to open-source projects of interest to Google. And given that R is widely used at Google, it's certainly a project of significant interest. So for the fifth year running the R Project will mentor students as part of GSOC 2012 to develop packages and improvements for R.

Tal Galili of R-bloggers provides a list of potential projects for students to work on, which include developing tools for open science with R, exporting R graphics as interactive applets for the Web, extending the knitR package to generate web-based documents, and much more. If you're a student with R and coding skills, this is a great way to improve the R ecosystem while earning some cash over the summer:

If you are a student looking to write some code for the R community, Google is willing to pay you $5000 for 3 months during this summer of 2012.  A good starting point is the R GSoC wiki. Interested students or mentors are encouraged to discuss other project ideas on the gsoc-r Google group.

For more information about R's involvement in GSOC 2012, see Tal's post below.

R-bloggers: Google summer of code 2012 – and R – a call for students

Video: Using R in Academic Finance

The slides and replay for Dr Sanjiv Das's webinar, Using R for Analyzing Loans, Portfolios and Risk: From Academic Theory to Financial Practice are now available. I've embedded the slides below: they tell a great story of how Das, after being mistaken for the then-CEO of Citibank (with whom he shares a name) was then led to research (using R software) into the reasons why many banks were not offering home-loan modifications despite the grave economic threat of foreclosures. He also invesitgates systemic risk in the financial system, and determines which banks really are "too big to fail". There are also sections on portfolio optimization with R, and resources for calling R from the Web.

The replay of the webcast is also available at the link below.

Revolution Analytics webinars: Using R for Analyzing Loans, Portfolios and Risk: From Academic Theory to Financial Practice

Revolution R Enterprise 5.0 now available for free academic download

Revolution R Enterprise 5.0, which we announced last week, is now available for free download to students and faculty at academic institutions worldwide. If you've downloaded Revolution R Enterprise via the academic program before and are on the mailing list, you will have already received an email with download instructions; if not, just complete the form linked below and download instructions will be emailed to you.

Note that tha academic program also includes free access to the RevoDeployR web services framework. If you have a project for integrating R into a web-based application or other third-party application, check the RevoDeployR distribution for the documentation to the server and client APIs.

Revolution Analytics: Download Revolution R Enterprise - Free Academic Program

10 reasons why a grad student should use R

Kevin Goulding is working on a Master’s degree in Applied Economics at Montana State University, and offers 10 reasons why grad students should choose R for statistical analyses, homework problems, and thesis research:

• R is free, and lets grad students escape the burdens of commercial license costs.
• R has really good online documentation; and the community is unparalleled.  
• The command-line interface is perfect for learning by doing. 
• R is on the cutting edge, and expanding rapidly.
• The R programming language is intuitive.  
• R creates stunning visuals. 
• R and LaTeX work together — seamlessly. 
• R is used by practitioners in a plethora of academic disciplines. 
• R makes you think.  
• There’s always more than one way to accomplish something. 

Kevin offers provides further details for each of these points in his blog post. Grad students can also expand their use of R with the additional features of Revolution R Enterprise free of charge.

The Tarzan: Why use R? A grad student’s 2 cents

 

Student travel grants for useR! 2011

For students planning to attend the annual worldwide R user conference, useR! 2011, travel grants are available to help defray the cost of attending the conference in the UK. CRISM is offering bursaries for accommodation and conference fees, and Revolution Analytics is offering $1,000 travel grants. Applications for funding, which must be accompanied by an abstract for a contributed talk or poster, must be submitted before March 11. Details at the useR! 2011 conference website, linked below.

useR! 2011: Funding