Saturday, 28 September 2013

Tableau 8 Guide

Recently released:  Tableau 8: The Official Guide   by George Peck

 I am in the midst of a project that specifies Tableau, and searched for several fairly obscure examples, and found them in this book.   Good sign.   Have used Tableau since soon after its release.   Tableau has an increasingly complex User Interface, because it covers lots of ground.  Will follow with a review as I read.  The book's site, with sample chapters.

" ... Present a unified view of complex BI using the entire Tableau 8 toolkit

Create and distribute dynamic, feature-rich data visualizations and highly interactive BI dashboards—quickly and easily! Tableau 8: The Official Guide provides the hands-on instruction and best practices you need to meet your business intelligence objectives and drive better decision making. Discover how to work from the Tableau GUI, load BI from disparate sources, drag and drop to analyze data, set up custom visualizations, and build robust dashboards. This practical guide shows you, step by step, how to design and publish meaningful business communications to end users across your enterprise. ... ? 

Thursday, 26 September 2013

Mapping Tutorials for Knowledge Capture

From Biggerplate, a set of tutorials on mind mapping.  Also related, from the IHMC site: Concept Maps, their theory and how to create them.   Examining for use in a knowledge capture problem.

Review: Modeling Techniques in Predictive Analytics


Modeling Techniques in Predictive Analytics: Business Problems and Solutions with R  by Thomas W. Miller 

This book does an excellent job of defining prediction, which I have rarely found done well  in a text.  The first chapter: Analytics and Data Science, in a relatively few pages, defines both the value of visualization and the methodology for getting and verifying a prediction.  Nicely and very simply and accessibly described.   The R code for this and every chapter can be found online here.  I have yet to use or test their code, but it seems very well documented.   This book does not teach R. I did not expect it to, ... but anyone who knows coding can follow along after getting an introduction.

Following chapters address specific areas of application:  Advertising and Promotion, Preference and Choice, Market Basket Analysis, Economic Data Analysis, Operations Management, Text Analytcs, Sentiment Analysis, Sports Analytics and Spatial Analysis.  Each of these topics contain a number of foundation problems, and a book of this type cannot cover them all, but the examples used are reasonable starting examples.

I have read the introductions of several of these topics, and the explanations are well done.  Code is printed on pages. and is often long and somewhat hard to read.  I will probably follow it on the online forms.

The text is heavily footnoted.   I like that, but the having the footnotes in the text make it somewhat harder to scan, would have preferred them in the index.

The Appendix contains description of a number of R language packs to support work described in the book.  Using a language like R is all about building on the tested work of others, so this is key. Ongoing work is described, but this will be out of  date quickly.  It may have been useful to have a place for social and formal exchanges and updates.  Think of the book as a nexus for social
interaction.

Overall,  is very well done,  I expect to use several of the examples for applications.
I have talked to two other people that have neeb reading and scanning it, and they also enjoyed its approach.

" ....  Thomas W. Miller, leader of Northwestern University’s pioneering program in predictive analytics, guides you through every step: defining problems, identifying data, crafting and optimizing models, writing effective R code, interpreting results, and more.
 Each chapter focuses on one of today’s most important applications for predictive analytics, giving you the skills and knowledge to put models to work–and gain maximum value from them. ... " 

Link Rot and Permanent Hyperlinks

We juggled with link rot since the earliest days of the Internet.  As we were trying to build from existing Internet content.  You can point to anything, but that thing can go away without warning.   There is a significant part of the Internet that has disappeared.  MJ Perry on the market for permanent hyperlinks.  We examined approaches like the Internet Archive for making content accessible.  For example, here is the archive for this blog up to May 30 of this year.  He covers some new explorations in this direction to make content selectively more permanent, like Perma.cc, soon to be made generally available and free.

Big Data Training from IBM and Universities

Educational Collaboration: 
 " ... IBM is hoping to help create the next generation of "big data" specialists through a series of partnerships with universities around the world, as well as influence the curriculum.
Nine new agreements announced Wednesday involve Georgetown University, George Washington University, Rensselaer Polytechnic Institute, the University of Missouri and Northwestern University in the U.S. IBM is also beginning big data programs at Dublin City University, Mother Teresa Women's University in India, the National University of Singapore, and the Philippines' Commission on Higher Education. ... " 

After the Design of Everyday Things

We interacted with designer Don Norman a few times, author of the ground breaking Design of Everyday Things.   This new interview in GigaOM relates well what he is up to.  Including a stealth startup about design in cooking.  A design MOOC to emerge in November.  Thoughts on alternative smartphone designs. And why skeumorphic design is being abandoned, and is that a good idea?   Will he use Google Glass?  Finally what is the most dangerous design out there?   Good reading.

Baby Name Voyager Visualization Again

Some time ago, for a problem we were working on for visualizing initiative data, we looked at the work of Martin Wattenberg.  Probably most famous is his Baby Name Voyager.  This takes social security data of the choice of baby names since the 1880s and cleverly visualizes the results into a highly interactive stacked line chart.  More technical information.  Although we did not use it for the problem at the time, it has come up again as an example for interacting with low dimension,  high volume data.  Worth taking another look, like I did.  Love to hear of other examples of its use.