BigML News, Issue #3, May 2013
BigML - Machine learning made easy

 

Hi <<Full Name>>,


We're excited to share several updates from BigML with you in this issue of our newsletter, including:
  • BigML for Minority Report Predictions
  • Featured Model: USA Life Expectancy
  • New features: Sunburst Visualizations & 1-Click Random Decision Forest
  • Machine Learning Explained: Ensembles
  • Call for BigML partners
  • Get yourself a free BigML subscription!
Customer showcase: BigML for Minority Report Predictions
Customer Identification & Personalization
CSC (Computer Sciences Corporation) is a global leader in providing technology-enabled business solutions and services. Like any business, CSC generates many customer leads from visitors to their website—leveraging data from Adobe Analytics as well as other tools and sources.  The person in charge of this activity for CSC is Christopher Marin, Director of Digital Marketing Platform at CSC.

In addition to the baseline information provided by Adobe Analytics, Chris used BigML to draw predictive analyses on CSC’s web visitors to determine which variables were most closely related to predicting a marketing lead.  By using BigML’s decision trees, CSC found key behaviors of likely customers that they otherwise wouldn’t have thought of as being critical—CSC can now incorporate this knowledge into their future marketing activities.
Christopher detailed his use of BigML as part of a recent presentation at Adobe Summit called "Minority Report comes of age: Predictive customer segmentation." 
Do you have a BigML deployment that you want to share with the world?  Let us know and if we write it up we’ll grant you some free modeling credits!
Featured Model: USA Life expectancy
USA Life expectancy Predictive Model
Quandl is a powerful tool that allows you to quickly pull data from a variety of public data sources and compile them into "supersets" that can then be shared and/or downloaded.  We've taken a superset created by a Quandl user that combines nearly 60 different fields, that we've used to predict life expectancy in the United States.
Check out the predictive model here, and then clone the model and dataset into your own account to start making your own models and predictions!
Latest feature: Sunburst Visualization
Sunburst visualization
While decision trees are in and of themselves very powerful for drawing insights from your data, BigML recently released support for Sunburst visualization of decision trees. This is very helpful for “needle in haystack” analyses and visualizations.

Read a guest blog post from David Gerster (Director of Data Science at Groupon) about how he sees value from this new functionality in BigML.
To use this new functionality for yourself, simply click on the sunburst icon when viewing your predictive model.
Machine Learning explained: Ensembles
Ensembles
It sounds counter-intuitive—if you already have a predictive model based on all of your data, why would multiple models based on subsets of your data be more accurate?  In short, running ensemble models protects against common machine learning pitfalls such as overfitting, and also better simulates your data in a future-looking manner (since your data is likely to change over time).  You can also think of running ensembles as getting additional opinions, and then making a final decision based on the outcomes of all of those opinions.  Read more about ensembles here and here.
Latest feature: 1-Click Random Decision Forest
Earlier this month BigML released support for Random Decision Forest predictive models within our interface.  Now you can have one-click access to the same powerful predictive algorithm that has been cited by Kaggle president Jeremy Howard as the foundation for many successful Kaggle competition entrants, and which we’ve tested to be superior to other machine learning techniques.
Predicting with ensembles
Become a BigML Partner
BigML is seeking complementary technology and consulting partners to help address a growing base of users and customer demand. If you are interested in partnering with BigML, please email us at partners@bigml.com
Big Opportunity:  Get A Free BigML Subscription
Recommend BigML
BigML is excited to bring a subscription option to our customers—this will allow you to use BigML as much as you want for a fixed monthly, quarterly or annual fee.  These paid subscriptions are coming around the corner, but in the interim we’re excited to invite you to get a free subscription for unlimited BigML usage, by simply inviting 3 or more friends to sign up for our service.  To get started, visit https://bigml.com/account/recommend.

Our mailing address is:
BigML, Inc
2851 NW 9th Street
Suite D, Conifer Plaza Building
Corvallis, Oregon 97330

Add us to your address book


Copyright © 2013 BigML, Inc, All rights reserved.