Skip to main content

Stand Back, I'm Going To Try Science!

Rachel Poulsen (Silicon Valley Data Science), John Akred (Silicon Valley Data Science)
Hadoop and Beyond
GA Ballroom J
Average rating: ***..
(3.43, 7 ratings)

Most are familiar with A/B testing of websites – a scientific method used to evaluate the effect of one factor by conducting a controlled experiment using a treatment site (with the new/changed factor) and a control site. A/B testing is just one type of experimental design – approaches used to test multiple factors and their interactions. In this session, we will walk through case studies to help participants understand how broadly they can apply experimental design principles throughout their projects, architectures and enterprises.

The term “Design of Experiments” refers to a systematic approach to understanding causality using data collection along with applied statistical techniques. Properly designed experiments control the environment of data collection, but this does not necessarily mean a carefully constructed lab in a clean building somewhere. In fact, most of us participate in designed experiments every day by simply browsing the web.

Applications of experimental design have come a long way in the last century. Its early industrial applications started in Agriculture in the 1930s. In the last several decades, it became widely popular in the manufacturing industry with the practice of Six Sigma, and has recently been used in website optimization.

In this session, we’ll review the “design” and the “experiment” side of Design of Experiments, from systematic data collection to basic statistical applications including experiments with multiple factors and discuss its future beyond the web. We will discuss examples ranging from understanding the performance impact of system improvements to understanding customer sensitivity to electricity prices.

Photo of Rachel Poulsen

Rachel Poulsen

Data Scientist, Silicon Valley Data Science

Rachel is a Data Scientist for Silicon Valley Data Science with a focus in Statistics and Communication. She has worked for American Express and TiVo and held the roles of Sr. Statistical Analyst and Analytics Product Manager. In her career, she has applied statistical techniques such as Statistical Process Control, Hierarchical and K-means Clustering, Principal Components, Logistic and Linear Regression, and Survival Analysis. At American Express, she used generalized additive models to optimize ROI for varying marketing campaigns. While at TiVo, she worked on projects benchmarking models for Ad Click Thru Rates, improving their Recommender system, analyzing patterns of user behavior, and building a monitoring system for measuring software responsiveness and detecting regressions. Rachel has a Master’s degree in Statistics, and a Bachelor’s in Mathematics and Communications.

Photo of John Akred

John Akred

CTO, Silicon Valley Data Science

With over 15 years in advanced analytical applications and architecture, John is dedicated to helping organizations become more data-driven. He combines deep expertise in analytics and data science with business acumen and dynamic engineering leadership.

Comments on this page are now closed.

Comments

Picture of Rachel Poulsen
Rachel Poulsen
02/12/2014 2:35pm PST

Hi all – slides will be available at strata.svds.com in the next week.