LinkedIn Skills Survey

I found LinkedIn Skills useful, but with many holes. Data mining "skill", while declined by 1% vs last year, greatly dominates other related skills, like Predictive Analytics or Web Mining

---

I recently came across LinkedIn Skills directories - turns out I have a top listing among Knowledge Discovery Professionals (~ 240 people), which is gratifying, since I coined the term "Knowledge Discovery in Data" and started the Knowledge Discovery in Data meetings in 1989.

However, while "Knowledge Discovery" remains popular among researchers, in the industry "Data Mining" has been much more popular, and since about 2006 (see Analytics surpassing Data Mining?) "Analytics" in various flavors, such as "Predictive Analytics".

Top LinkedIn Data Mining Professionals are

• Ronny Kohavi, Partner Architect at Microsoft
• Dean Abbott, Data Mining and Predictive Analytics Expert
• Karl Rexer, Data Mining & Analytic CRM Consulting
• Rob Cooley, CTO at OptiMine Software
• Andreas Weigend, People and Data / Weigend Associates LLC
• Gregory Piatetsky-Shapiro, Analytics/Data Mining Consultant; Editor, KDnuggets; Founder KDD/SIGKDD
• Michael Berry, Founder and Principal, Data Miners, Inc.
• Robert Grossman, Professor at the University of Chicago and Partner at Open Data Group

It is not clear how the ranking is determined, but it is not only by number of connections, since #2 on that list, Dean Abbott, has about 40% fewer connections than #3, Karl Rexer.

While "data mining" growth has been -1% vs last year (compared to +5% for "web mining" skill), the "size" of data mining (I assume it is the number of people with that skill) is about 22,000, much more than for related fields like "Predictive Modeling" (2,500), "Text Mining" (~1,800), "Web Mining" (320), and "Predictive Analytics" (~ 3,000).

Top Predictive Analytics Professionals are

• Dean Abbott, Data Mining and Predictive Analytics Expert
• Karl Rexer, Data Mining & Analytic CRM Consulting
• Khosrow Hassibi, Principal Solutions Architect at SAS Institute Inc
• Jason Verlen, Director, Product Strategy & Product Management, Predictive Analytics at IBM
• Iqbal Kaur, Analytics & Reporting Leader at Target
• Vincent Granville, Chief Scientist at LookSmart
• James Taylor, Leading authority on Decision Management. Independent analyst/consultant in business rules and predictive analytics
• Tom Khabaza, Predictive Analytics & Data Mining

The lists of "related" fields look mostly plausible, but there are glaring omissions, for example "Predictive Analytics" or "Knowledge Discovery" are not related to "Data Mining", according to LinkedIn. Is it because that people do not list both on their profiles?

What do you think about LinkedIn Skills?

Comments:

Peter Skomoroch
Gregory,
Glad to see you showed up for data mining, well deserved. The list of related skills are currently based on co-occurence of phrases in member profiles, along with some corrections for relevance. LinkedIn Skills is still in Beta, stay tuned for algorithmic improvements in the "related skills", PeopleRank, and disambiguation.

-Pete
(Lead Data Scientist on LinkedIn Skills)

Gregory Piatetsky
Pete, thank you for a quick feedback! Lots of interesting opportunities in your data. Does "skill size" correspond to number of people with that skill?

Gregory

Peter Skomoroch
Yes, the skill size corresponds to the number of member profiles we have tagged with that skill. The skill tagger is set to be a bit conservative at the moment, so these numbers are likely an underestimate of the total population on LinkedIn.

For example, there are 4200 results for Hadoop in LinkedIn Search:

www.linkedin.com/search/fpsearch?type=people&keywords=hadoop

On the skill page for Hadoop, the size is < 2000:

www.linkedin.com/skills/skill/Hadoop?trk=tyah

Tom H. C. Anderson
Gregory, I think this may very well be connected tot he Mixtent application I wrote about on my blog last week. Quite interesting:

www.tomhcanderson.com/2011/04/06/why-dont-market-research-firms-get-social-media/

Tom

Gregory Piatetsky
Tom,
Mixtent is an interesting application, but it wants users to rank otehr people in their network. Ranking "hot" or "not" is fine in dating scene, but I think many professionals (me included) are not comfortable with it. Or is it just generation thing - younger people will not have a problem with such ranking?

LinkedIn was until recently much less active than facebook in its use of social graph, but I expect we will soon see a lot more social applications on its platform.

Victor Soroka
Hi Gregory!
Probably you will be interested in updated rating of LinkedIn skills
italabre.blogspot.com/2011/12/first-global-rating-of-linkedin-skills.html

Gregory PS
Victor,
thanks for the link - I will redo the Analytics-related skill analysis in 2012