Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
Data science : create teams that ask the right questions and deliver real value
Rose D., Apress, New York, NY, 2016. 251 pp. Type: Book (978-1-484222-52-2)
Date Reviewed: Jun 8 2017

When building a data science team, special care must be dedicated to its composition for organizations to obtain the best possible benefits. A data science team is a special kind of animal in the corporate zoo because it encompasses roles and activities uncommon to the classical business structure and operation.

This fine book (albeit a little too lengthy at times) guides those responsible for building such a team on how to build the best possible data science team for her/his organization. The author explains that the focus of such a team is not technology but business, is not algorithms but asking the right business-relevant questions that are answerable by the data the organization collects (correct data collection is also the data science team’s responsibility). Obviously, technology and algorithms are the enablers of the data-science-related revolution that businesses and organizations witness these days. But the focus of data science teams is not computers, is not statistics, is not big data analytics; it is the optimum use of these tools to obtain results and insights that are relevant to the organization, and the optimum communication of these results within the organization.

To successfully accomplish their mandate, a data science team must, according to the author, have a special structure that must contain the following roles: research lead, data analyst, and project manager. A research lead is a person that understands the business, but also has the ability to think outside the box about the business. She can identify assumptions and constraints, can unravel the interesting questions, and in a sense has the capacity to separate the data from the questions. She is not interested in the “how” to get an answer, or how to make the data provide an answer, but on the “why” something happens, and also, “what” might happen if such and such a thing were true, and so on; that is, she is not afraid to experiment with “well-established” notions or “common” assumptions, and she questions “common” ideas and practices (that are hidden in the data).

The data analyst is the one who knows technologies and algorithms, statistics and analytics, databases (relational and not relational), and machine learning. It is the role that collects the data, cleans and normalizes the data, makes sense of the data by selecting and using the appropriate tools for each situation, and creates reports that help with the discovery of interesting points and decision-making. These people, as the author says, usually come mainly from computer science and not so much from statistics; they have a multidisciplinary nature in their thinking and skill set and also know some software development.

The team must also have a project manager to keep it on track. This role, apart from governing the portfolio of all parallel projects that the team is running in a classical project manager sense, is also responsible for communicating the results and enforcing organizational learning. Also, the project manager shields the team to protect it from numerous senseless meetings and is at the same time a facilitator in creating those special communication links that the team needs when tackling a novel issue--links that are usually absent in large classically structured organizations where many semi-isolated silos of structure, activities, and most importantly data exist.

To make a long story short, this book is mostly about the sociology of data science teams and not so much about data science. It explains not only the internal workings of optimum data science teams, but also the activities that happen around such teams. The author gives a lot of field stories and examples to demonstrate his ideas, and takes special care to emphasize the bad practices that have haunted the first efforts of data science team creations and how to avoid them.

This is a very useful book for managers who need to create and govern teams that handle data science in the best possible way for their organizations, without having to learn data science.

More reviews about this item: Amazon

Reviewer:  Constantin S. Chassapis Review #: CR145337 (1708-0527)
Bookmark and Share
 
General (K.6.0 )
 
 
Business (J.1 ... )
 
 
General (H.3.0 )
 
Would you recommend this review?
yes
no
Other reviews under "General": Date
Disaster recovery handbook
, TAB Books, Blue Ridge Summit, PA, 1991. Type: Book (9780830676637)
Aug 1 1991
Management of information technology
Frenzel C., Boyd & Fraser Publishing Co., Danvers, MA, 1992. Type: Book (9780878355082)
Sep 1 1992
Production software that works
Behuniak J., Ahmad I., Courtright A., Digital Press, Newton, MA, 1992. Type: Book (9781555580834)
Dec 1 1992
more...

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy