Top LinkedIn Groups for Analytics, Big Data, Data Mining, and Data Science – Discussions up, Engagement down
While discussions are growing, the comments and engagements are falling, especially since 2012. We cluster groups into 4 quadrants by activity level and identify most active and engaged groups. Open groups are twice as active as closed.
We continue our
analysis of Top LinkedIn Groups for Analytics, Big Data, Data Mining, and Data Science. Last month we examined growth from groups "Big Bang" in 2008 to present and in this part we look at activity - comments, discussions, and engagement.
Our key findings
Next chart shows comments and discussions per group, for each 12 months period, from Q2 2009 to Q1 2015.
Fig. 1: Top LinkedIn Analytics, Big Data, Data Mining, Data Science Groups,
Comments & Discussions per year, 2009 to 2015
We note that while the total number of discussions is growing, the total number of comments actually started to decline in 2013, despite the growth in membership. The big gap between the average and the median values shows the wide range in activity levels between the groups.
We can see the trends more clearly if we measure the comments and discussions per week and per 1000 members.
Fig. 2: Top LinkedIn Analytics, Big Data, Data Mining, Data Science Groups,
Comments & Discussions per week per 1000 members
Note that LinkedIn group statistics only give discussion counts starting around June 2010, while comment counts are available starting from Sep 2008.
The discussion numbers per member were growing and peaked in 2012, with the launch of 2nd cluster of Big Data groups in 2012 (some of them were very active), but both discussions and activity levels are declining after 2012.
An important factor is group openness. 20 of the top 35 groups are open, and open groups have over twice as many comments (median 0.59) and discussions (median 1.61) as closed groups.
Fig. 3: Top LinkedIn Analytics, Big Data, Data Mining, Data Science Groups,
Open vs Closed groups Comments, Discussions/week per 1000 members, 14Q2 to 15Q1
Here are the 10 groups with the most comments per 1000 members per week in 12 months from 2014 Q2 to 2015 Q1.
Groups with most comments/week per 1000 members
Graphing comments levels for all groups shows an overall decline in levels of comments.
And here are the 10 groups with the highest discussions levels per 1000 members per week in 12 months from 2014 Q2 to 2015 Q1.
Groups with most discussions/week per 1000 members
Next graph shows the wild nature of discussions for different groups, going as high as 20/week per 1k members in 2012 for BDAHN (Big Data, Analytics, Hadoop, NoSQL & Cloud Computing) group, but declining for most groups. See all group abbreviations in the table at the end of this post.
Fig. 4: Top LinkedIn Analytics, Big Data, Data Mining, Data Science Groups,
Discussions per week per 1000 members, 2009 to 2015
We can measure group member engagement as the number of comments per week divided by the number of discussions per week (note that per 1000 member factor cancels out).
Fig. 5: Top LinkedIn Analytics, Big Data, Data Mining, Data Science Groups,
Average Engagement per week
Overall, the average engagement across all groups has declined two-fold from 2.16 in Q2 2010 to 1.03 in Q1 2015, and median engagement declined six-fold from 1.33 to 0.21. This suggest a number of very active groups which stand out.
Here are the 10 groups with the highest engagement levels in in 12 months from 2014 Q2 to 2015 Q1. We note however that groups with high engagement levels mostly have low discussion levels
Groups with highest engagement (cmts/wk per 1k mbr / disc/wk per 1k mbr)
Next chart shows shows group comments/week per 1,000 members vs group discussions/week per 1,000 members. Group name abbreviations are in the table below. Since the distributions of comments and discussions are very asymmetrical, we use log scale axes.
The median along each axis create 4 quadrants: Active, Commenting, Posting, and Passive.
Fig. 6: Top LinkedIn Analytics, Big Data, Data Mining, Data Science Groups,
Activity, 14Q2 to 15Q1, Comments/week vs discussions/week per 1000 members
Closed groups are squares while open groups are circles.
4 Quadrants: Active, Commenting, Posting, and Passive
For better separation of groups, both axes are on logarithmic scale
The median lines on each dimension create 4 quadrants:
Here are the groups in each quadrant, in order of activity = comments/wk per 1k members + discussions/wk per 1k members
Active: both discussions and comments above median:
Commenting, bottom right - discussions below median but comments above median
Posting: comments below median, discussions/posts above median:
Passive (both comments and posts below median):
The details are in the table with below. This table orders the groups by activity. Since the distribution of comments and discussions is very uneven, we computed separate "comment" rank and "discussion" rank for each group, and order groups by Avg Rank - their average rank. Eg KDnuggets is n. 1 in discussions/1k members and n. 3 in comments/1k members so its average rank is 2. In case of ties, we choose the group with highest sum of comments and discussions.
A group is open, unless it has a lock icon
Table 1: Top LinkedIn Analytics, Big Data, Data Mining, and Data Science groups,
Comments and Discussions, 2014 Q2 - 2015 Q1.
Values 25% or more higher than median are in green,
25% or more lower than median are in red, and the rest are in black.
Related:
Our key findings
- as groups grow, discussions increase, comments decline - surprisingly even in absolute numbers (!)
- engagement (comments/discussions) slows down
- Open groups are twice as active, in both comments and discussions
- We identify 4 group quadrants (see below): Active, Commenting, Posting, Passive
- Most active groups: KDnuggets,
Data Scientists,
Data Science & Machine Learning,
Big Data and Analytics,
RDataMining
Next chart shows comments and discussions per group, for each 12 months period, from Q2 2009 to Q1 2015.
Fig. 1: Top LinkedIn Analytics, Big Data, Data Mining, Data Science Groups,
Comments & Discussions per year, 2009 to 2015
We note that while the total number of discussions is growing, the total number of comments actually started to decline in 2013, despite the growth in membership. The big gap between the average and the median values shows the wide range in activity levels between the groups.
We can see the trends more clearly if we measure the comments and discussions per week and per 1000 members.
Fig. 2: Top LinkedIn Analytics, Big Data, Data Mining, Data Science Groups,
Comments & Discussions per week per 1000 members
Note that LinkedIn group statistics only give discussion counts starting around June 2010, while comment counts are available starting from Sep 2008.
The discussion numbers per member were growing and peaked in 2012, with the launch of 2nd cluster of Big Data groups in 2012 (some of them were very active), but both discussions and activity levels are declining after 2012.
An important factor is group openness. 20 of the top 35 groups are open, and open groups have over twice as many comments (median 0.59) and discussions (median 1.61) as closed groups.
Fig. 3: Top LinkedIn Analytics, Big Data, Data Mining, Data Science Groups,
Open vs Closed groups Comments, Discussions/week per 1000 members, 14Q2 to 15Q1
Here are the 10 groups with the most comments per 1000 members per week in 12 months from 2014 Q2 to 2015 Q1.
Groups with most comments/week per 1000 members
- RDataMining: R and Data Mining, open, 2.31
- Data Scientists, open, 1.38
- KDnuggets Analytics, Data Mining, and Data Science, open, 1.34
- Big Data, Analytics, Hadoop, NoSQL & Cloud Computing, open, 1.24
- Statistics & Analytics Consultants Group, (closed), 1.02
- Advanced Business Analytics, Data Mining and Predictive Mode, open, 0.83
- Big Data and Analytics, closed, 0.80
- Next Gen Market Research (NGMR), open, 0.80
- Text Analytics, open, 0.76
- Data Science & Machine Learning, open, 0.74
Graphing comments levels for all groups shows an overall decline in levels of comments.
And here are the 10 groups with the highest discussions levels per 1000 members per week in 12 months from 2014 Q2 to 2015 Q1.
Groups with most discussions/week per 1000 members
- KDnuggets Analytics, Data Mining, and Data Science, open, 8.86
- Data Scientists, open, 6.66
- IBM Big Data and Analytics, open, 4.54
- Data Science & Machine Learning, open, 4.39
- Predictive Analytics Network , open, 3.61
- BIG DATA Professionals - Architects Scientists IOT Analytics, open, 3.41
- Data & Text Analytics Professionals , open, 3.21
- Big Data and Analytics, closed, 2.86
- Data Mining, Statistics, Big Data, and Data Visualization, open, 2.74
- Business Analytics, open, 2.48
Next graph shows the wild nature of discussions for different groups, going as high as 20/week per 1k members in 2012 for BDAHN (Big Data, Analytics, Hadoop, NoSQL & Cloud Computing) group, but declining for most groups. See all group abbreviations in the table at the end of this post.
Fig. 4: Top LinkedIn Analytics, Big Data, Data Mining, Data Science Groups,
Discussions per week per 1000 members, 2009 to 2015
We can measure group member engagement as the number of comments per week divided by the number of discussions per week (note that per 1000 member factor cancels out).
Fig. 5: Top LinkedIn Analytics, Big Data, Data Mining, Data Science Groups,
Average Engagement per week
Overall, the average engagement across all groups has declined two-fold from 2.16 in Q2 2010 to 1.03 in Q1 2015, and median engagement declined six-fold from 1.33 to 0.21. This suggest a number of very active groups which stand out.
Here are the 10 groups with the highest engagement levels in in 12 months from 2014 Q2 to 2015 Q1. We note however that groups with high engagement levels mostly have low discussion levels
Groups with highest engagement (cmts/wk per 1k mbr / disc/wk per 1k mbr)
- Business Intelligence Professionals , 11.03 (0.23 / 0.02)
- Pattern Recognition, Data Mining, Machine Intelligence, 6.93 (0.67 / 0.10)
- Statistics & Analytics Consultants Group, 3.04 (1.02 / 0.34)
- Actuary / Actuarial, Predictive Modeling, Data Mining, 2.04 (0.14 / 0.07)
- Advanced Analytics, Predictive Modeling & Statistical, 1.96 (0.48 / 0.25)
- Big Data, Analytics, Hadoop, NoSQL & Cloud Computing, 1.72 (1.24 / 0.72)
- RDataMining: R and Data Mining, 1.26 (2.31 / 1.84)
- Text Analytics, 1.02 (0.76 / 0.75)
- Machine Learning Connection, 0.93 (0.58 / 0.63)
- Advanced Business Analytics, Data Mining and Predictive Mode, 0.84 (0.83 / 0.99)
Next chart shows shows group comments/week per 1,000 members vs group discussions/week per 1,000 members. Group name abbreviations are in the table below. Since the distributions of comments and discussions are very asymmetrical, we use log scale axes.
The median along each axis create 4 quadrants: Active, Commenting, Posting, and Passive.
Fig. 6: Top LinkedIn Analytics, Big Data, Data Mining, Data Science Groups,
Activity, 14Q2 to 15Q1, Comments/week vs discussions/week per 1000 members
Closed groups are squares while open groups are circles.
4 Quadrants: Active, Commenting, Posting, and Passive
For better separation of groups, both axes are on logarithmic scale
The median lines on each dimension create 4 quadrants:
- Active, top right - both discussions and comments above median
- Commenting, bottom right - discussions below but comments above median
- Posting: top left, discussions above, but comments below median
- Passive: bottom left, both discussions and comments below median.
Here are the groups in each quadrant, in order of activity = comments/wk per 1k members + discussions/wk per 1k members
Active: both discussions and comments above median:
- KDnuggets (KDnuggets), Activity=10.20
- Data Scientists ((DScient)) , 8.04
- Data Science & Machine Lea (ning (DS) & ML), 5.13
- Predictive Analytics Network (PAN) , 4.24
- RDataMining: R and Data Mining (RDM), 4.15
- BIG DATA Professionals - Architects Scientists IOT Analytics (BD pros), 4.07
- Big Data and Analytics (BD & A), 3.66
- Data Mining, Statistics, Big Data, and Data Visualization (DMSBD), 3.28
- Machine Learning and Data Science (ML/DSC), 2.66
Commenting, bottom right - discussions below median but comments above median
- Next Gen Market Research (NGMR), 1.96
- Big Data, Analytics, Hadoop, NoSQL & Cloud Computing (BDAHN), 1.96
- Advanced Business Analytics, Data Mining and Predictive Mode (Adv BADM), 1.81
- Text Analytics, 1.52
- Statistics & Analytics Consultants Group, 1.36
- Machine Learning Connection, 1.21
- Pattern Recognition, Data Mining, Machine Intelligence ..., 0.77
- Advanced Analytics, Predictive Modeling & Statistical ..., 0.73
Posting: comments below median, discussions/posts above median:
- IBM Big Data and Analytics, 4.92
- Data & Text Analytics Professionals , 3.37
- Advanced Analytics, 2.91
- Business Analytics, 2.89
- Global Analytics Network , 2.25
- Research Methods and Data Science, 1.77
- Predictive Analytics, 1.62
- Business Intelligence & Analytics Group, 1.50
Passive (both comments and posts below median):
- Visual Analytics, 1.43
- Healthcare Data Mining and Modeling, 1.41
- Big Data | Analytics | Strategy | Finance | Innovation, 1.39
- Data Mining Technology, 1.35
- SAS & Analytics Users, 1.28
- Data Warehouse / Big Data / Hadoop / Predictive Analytics, 0.88
- SAS Analytics & BI, 0.54
- Business Intelligence Professionals , 0.25
- Actuary / Actuarial, Predictive Modeling, Data Mining ..., 0.20
- Lavastorm Analytics Community Group, 0.20
The details are in the table with below. This table orders the groups by activity. Since the distribution of comments and discussions is very uneven, we computed separate "comment" rank and "discussion" rank for each group, and order groups by Avg Rank - their average rank. Eg KDnuggets is n. 1 in discussions/1k members and n. 3 in comments/1k members so its average rank is 2. In case of ties, we choose the group with highest sum of comments and discussions.
A group is open, unless it has a lock icon
Table 1: Top LinkedIn Analytics, Big Data, Data Mining, and Data Science groups,
Comments and Discussions, 2014 Q2 - 2015 Q1.
Values 25% or more higher than median are in green,
25% or more lower than median are in red, and the rest are in black.
| LinkedIn Group | Avg Rank / 1k mbr |
Cmts/wk / 1k mbr |
Disc/wk / 1k mbr |
Members (Mar 30, 2015) |
|---|---|---|---|---|
| Median | na | 0.451 | 1.286 | 18377 |
KDnuggets Analytics, Data Mining, and Data Science (KDnuggets)
owner: Gregory Piatetsky-Shapiro |
2 | 1.34 | 8.86 | 8560 |
Data Scientists (DScient)
owner: Troy Sadkowsky |
2 | 1.38 | 6.66 | 14063 |
Data Science & Machine Learning (DS & ML)
owner: Pavandeep Kalra |
7 | 0.74 | 4.39 | 8895 |
RDataMining: R and Data Mining (RDM)
owner: Yanchang Zhao |
7.5 | 2.31 | 1.84 | 11677 |
Big Data and Analytics (BD & A)owner: Sarah Howes |
7.5 | 0.8 | 2.86 | 120776 |
Predictive Analytics Network (PAN)
owner: Srikanth Velamakanni |
9 | 0.63 | 3.61 | 12604 |
BIG DATA Professionals - Architects Scientists IOT Analytics (BD pros)
owner: Qamar Zia |
9 | 0.66 | 3.41 | 45443 |
IBM Big Data and Analytics (IBM BDA)
owner: Bruce Weed |
12 | 0.38 | 4.54 | 14277 |
Data Mining, Statistics, Big Data, and Data Visualization (DMSBD)
owner: Jon Francis |
12 | 0.54 | 2.74 | 68539 |
Next Gen Market Research (NGMR) (NGMR)
owner: Tom HC Anderson |
14 | 0.8 | 1.17 | 24049 |
Machine Learning and Data Science (former DSC) (ML/DSC)owner: Richard Snee |
14.5 | 0.46 | 2.2 | 20876 |
Advanced Analytics (Adv An)owner: Osris Arya |
15 | 0.43 | 2.48 | 16453 |
Business Analytics (Biz An)
owner: Alberto Roldan |
15 | 0.41 | 2.48 | 70326 |
Advanced Business Analytics, Data Mining and Predictive Mode (Adv BADM)
owner: Vincent Granville |
15 | 0.83 | 0.99 | 174362 |
Big Data, Analytics, Hadoop, NoSQL & Cloud Computing (BDAHN)
owner: VenkataHari Shankar |
15.5 | 1.24 | 0.72 | 33139 |
Research Methods and Data Science (RMDS)
owner: Alex Liu |
17 | 0.45 | 1.32 | 18377 |
Text Analytics (Text A)
owner: Maria Milosavljevic |
17.5 | 0.76 | 0.75 | 16404 |
Statistics & Analytics Consultants Group (SAC)owner: Burke Powers |
17.5 | 1.02 | 0.34 | 46481 |
Data & Text Analytics Professionals (D&TA Prof)
owner: Tom HC Anderson |
18 | 0.16 | 3.21 | 9086 |
Predictive Analytics (Pred An)
owner: William Erwin |
19.5 | 0.24 | 1.38 | 6945 |
Machine Learning Connection (ML Conn)owner: Shane Threatt |
21 | 0.58 | 0.63 | 24712 |
Business Intelligence & Analytics Group (BI&A)
owner: Rakesh Rajora |
22 | 0.21 | 1.29 | 20000 |
Visual Analytics (Visual)
owner: Christian Posse |
22 | 0.24 | 1.18 | 10762 |
Pattern Recognition, Data Mining, Machine Intelligence ... (PRDM)owner: Belur V. Dasarathy |
22 | 0.67 | 0.1 | 21813 |
Healthcare Data Mining and Modeling (Healthcar)
owner: Raghu Santhanam |
22.5 | 0.35 | 1.06 | 4684 |
Big Data | Analytics | Strategy | Finance | Innovation (BDASFI)owner: Josie King |
22.5 | 0.25 | 1.13 | 128805 |
Global Analytics Network (Global A)owner: Roni Lynn Zapin |
23 | 0.08 | 2.16 | 20432 |
Advanced Analytics, Predictive Modeling & Statistical ... (Adv AP)owner: John Frischenmeyer |
23.5 | 0.48 | 0.25 | 9185 |
Data Mining Technology (DMT)owner: Stan Byrne |
26 | 0.07 | 1.29 | 4854 |
SAS & Analytics Users (SAS Users)owner: Michael Kaushansky |
26 | 0.14 | 1.15 | 18283 |
Data Warehouse / Big Data / Hadoop / Predictive Analytics (DW/BD/H)owner: Chris Rosser |
28.5 | 0.12 | 0.75 | 32994 |
SAS Analytics & BI (SAS A&BI)owner: Stan Byrne |
28.5 | 0.18 | 0.36 | 24575 |
Business Intelligence Professionals (BI Pros)owner: Rajasekar Nonburaj |
30.5 | 0.23 | 0.02 | 150147 |
Actuary / Actuarial, Predictive Modeling, Data Mining ... (Actuary)owner: Tom Troceen |
32 | 0.14 | 0.07 | 17616 |
Lavastorm Analytics Community Group (Lavastorm)
owner: William Thomas |
33.5 | 0.03 | 0.17 | 7126 |
Related:
- Top LinkedIn Groups for Analytics, Big Data, Data Mining, and Data Science – from “Big Bang” to Now, Apr 2015.
- Update: Top LinkedIn Groups for Analytics, Big Data, Data Mining, and Data Science, Nov 2015.
- Top LinkedIn Groups in 2014 for Analytics, Big Data, Data Mining, and Data Science, Apr 2014.
- Top 2013 LinkedIn Groups for Analytics, Big Data, Data Mining, and Data Science, Dec 2013.
- Top LinkedIn Groups for Analytics, Big Data, Data Mining, and Data Science, Apr 2013.
