"Official" February 2014 BARUG Meetup

Agenda:
6:30 - Networking
6:50 - Announcements
6:55 - Tom Johnson: Cloud-based Shiny App Calculates Live Bitcoin Arbitrage in Browser
7:25: - Hadley Wickham: Automated testing in R with testthat, R CMD check and travis-ci
8:00 - Michael Helbraun: Modeling churn with Hadoop, Revolution R Enterprise and Tableau

Tom Johnson:
Abstract
Priceonomics.com invited people to solve a puzzle to find, at each point in time, the best 2-currency plus Bitcoin arbitrage among 3 currencies, e.g., USD, EUR, JPY. They provided a live feed of synthetic exchange rates, updated every second.

Using R for about everything, I wrote a script to determine the most profitable sequence of transactions from any chosen home currency, through an intermediate currency, then through Bitcoin and finally back to the home currency. I then embedded it in a Shiny app. While it can run on any R platform, I will demo it implemented as a website, hosted, in the AWS cloud.

This app is only a demo, as it does not use real, executable prices. I am also extending the approach beyond calculating and displaying data, to monitoring and controlling live systems. I find R and Shiny together free me from many development complexities.

About
Tom is an independent consultant with his bay area company Fortunet, Inc., developing quantitative financial management systems. He now focuses on real-time decisions-and-execution based on detection of patterns in non-stationary data-streams. Latest is a fully automated foreign currency trading system (generating signals through executing trades). As a Registered Commodity Trading Adviser, he developed and applied valuation models, exposure management rules, investment performance tools for banks and investment firms. Tom has also been a principal in start-ups, corporate management, and done organization development consulting.

This range of experience draws on his interdisciplinary PhD from Stanford University which focused on the Valuation of Human Capital within the Firm (Business, Engineering, Economics, Education and Statistics). Before that, Graduate Fellow at RAND Graduate Institute, MS in Management from University of California at Irvine (Top graduate), Captain in US Marine Corps, BA at University of Denver with Honors in Psychology.

Hadley Wickham

Hadley is the creator of many popular R packages including ggplot2, plyr, reshape2, and devtools and dplyr.

Michael Helbraun
Abstract
A typical data science project involves, end users and business analysts as well as data scientists, Michael will illustrate the process with a customer churn application showing how the statistical algorithms Revolution Analytics RevoScaleR R package run inside a Hadoop cluster to analyze large data sets, and how the results obtained may be displayed with Tableau 8.1 Software.

Michael is a Senior Sales Engineer with Revolution Analytics