Machine Learning Newsletter

2200 CS Faculty in 51 University Analysis

Data is from here and explanation of how it is collected is here. It was a classroom assignment to crowdsource the faculty information(when they join to the university, what is their rank and where they get doctorate and so on). The not-so good taxonomy for the fields that faculty are working is taken from Microsoft Academic Search. With data integrity caveats in mind, let's look at the data more closely.

I made an interactive visualization, if you want to explore the dataset without thinking about the temporal domain too much, play with it.

I am mainly interested how things vary over time in this post, but the categories(subfields as well as the number of universities) are hard to put into a single graph, so I separated into different sections. Even I divided the number of universities based on the number of professorship.

Let's dive in right a way.

In [4]:
 
Out[4]:
Name University JoinYear Rank Subfield Bachelors Masters Doctorate PostDoc
0 Assaf J. Kfoury Boston University NaN Full Programming Languages Massachusetts Institute of Technology - USA Massachusetts Institute of Technology - USA Massachusetts Institute of Technology - USA NaN
1 Azer Bestavros Boston University 1991 Full Networks & Communications Alexandria University - Egypt Alexandria University - Egypt Harvard University - USA NaN
2 Evimaria Terzi Boston University 2009 Assistant Data Mining Aristotle University of Thessaloniki - Greece Purdue University - USA University of Helsinki - Finland Helsinki Institute for Information Technology ...
3 George Kollios Boston University 2000 Full Databases National Technical University of Athens - Greece Polytechnic University - Brooklyn - New York -... Polytechnic University - Brooklyn - New York -... NaN
4 Hongwei Xi Boston University 2001 Associate Programming Languages Nanjing University - China Nanjing University - China Carnegie Mellon University - USA NaN

5 rows × 9 columns

The data looks like as above, we have the faculty name, join year, rank in the university, Bachelors, Masters, Doctorate, PostDoc universities. Generally, JoinYear, Rank, Subfield are more complete and the universitites in the Bachelors, Masters and Doctorate are less complete. The most sparse is the PostDoc field.

In [31]:
 
stat_bin: binwidth defaulted to range/30.
    Use 'binwidth = x' to adjust this.
Out[31]:
<ggplot: (278616465)>

This graph can be explained in two ways. First the data collection may favor the recent, new hires as they may have more likely to have updated information, webpages and so on. Consider a professor's web existence who joined to the university at 1970 vs an assistant professor who just joined to a university. When we look at the average in 80's and 90's we would see a consistent line whereas when we come to 2000's, we see a quite drastic jump. As software is eating world, this would signal from 2000 to become such fields. What about future? Most of the projections regarding CS seems promising for field and signals even more growth.

In [32]:
 
Out[32]:
<ggplot: (275289897)>

Full rank professors are a lot as expected. Apart from the age distribution of the professors, full suggests tenure and it is the longest period of the faculty's academic career. Moreover, assistant period is limited for 6-7 years, if the professor cannot get the tenure, s/he may move to another university or to industry.

Top 10 Fields by Hiring Number

I think the most problematic part of this dataset is the taxonomy of the fields in computer science. That being said, it is still surprising to see the Algorithms & Theory is the largest field by far. It is almost same with the sum of second (Network & Communications) and third (Hardware & Architecture) largest fields.

In [7]:
 
Out[7]:
Name University JoinYear Rank Subfield Bachelors Masters Doctorate PostDoc
Subfield
Algorithms & Theory 293 293 232 293 293 223 169 293 78
Networks & Communications 163 163 135 163 163 140 127 163 20
Hardware & Architecture 150 150 119 150 150 118 99 150 12
Artificial Intelligence 147 147 119 147 147 121 100 147 30
Bioinformatics & Computational Biology 131 131 105 131 131 103 73 131 39
Human-Computer Interaction 127 127 103 127 127 105 80 125 14
Security & Privacy 121 121 101 121 121 106 89 120 22
Graphics 113 113 84 113 113 92 82 113 17
Programming Languages 99 99 77 99 99 77 65 98 11
Machine Learning & Pattern Recognition 95 95 79 95 95 73 63 95 29
Distributed & Parallel Computing 93 93 73 93 93 75 61 93 13
Scientific Computing 90 90 64 90 90 68 58 90 14
Computer Vision 90 90 72 90 90 71 59 90 21
Software Engineering 83 83 64 83 83 65 51 83 9
Databases 77 77 66 77 77 59 54 76 7
Operating Systems 72 72 53 72 72 63 53 71 6
Natural Language & Speech 51 51 41 51 51 42 32 51 16
Computer Education 49 49 36 49 49 40 29 48 10
Real-Time & Embedded Systems 32 32 27 32 32 31 26 32 4
Data Mining 31 31 26 31 31 24 20 31 5
Multimedia 15 15 11 15 15 10 12 14 1
Information Retrieval 6 6 6 6 6 6 6 6 0
World Wide Web 5 5 4 5 5 4 3 5 0

23 rows × 9 columns

In [11]:
 
Out[11]:
<ggplot: (278625041)>

Algorithms & Theory are consistently higher for most of the years where Human Computer Interaction gets popular in the recent years. Once upon a time, Hardware & Architecture used to be popular. Another field that gained traction is Machine Learning & Pattern Recognition.

In [13]:
 
Out[13]:
<ggplot: (278625717)>

In here Software Engineering is consistently high, Computer Vision and Distributed & Parellel computing gained traction in the recent years.

What ranks in professorship increases over time?

In [14]:
 
Out[14]:
Name University JoinYear Rank Subfield Bachelors Masters Doctorate PostDoc
Rank
Full 1209 1209 931 1209 1209 947 797 1205 127
Associate 543 543 409 543 543 441 367 541 88
Assistant 355 355 339 355 355 304 227 355 160
Adjunct 8 8 4 8 8 8 7 8 2
Research 6 6 4 6 6 6 3 5 1
Practice 5 5 5 5 5 4 4 4 0
Affiliate 4 4 3 4 4 3 3 4 0
Teaching 1 1 0 1 1 1 1 1 0
Lecturer 1 1 1 1 1 1 1 1 0
Clinical 1 1 1 1 1 1 1 1 0

10 rows × 9 columns

In [16]:
 
Out[16]:
<ggplot: (279356157)>

If you look at the full time professsors in recent years, you would see the transfers of academics between universities which is somehow low than I expected. Interesting thing is that sum of full rank professors and associate professors seem to be constant as if you could see the transition of associate to full rank.

In [18]:
 
Out[18]:
Name University JoinYear Rank Subfield Bachelors Masters Doctorate PostDoc
Doctorate
Massachusetts Institute of Technology - USA 251 251 198 251 251 218 186 251 52
University of California - Berkeley - USA 179 179 153 179 179 155 110 179 33
Stanford University - USA 148 148 120 148 148 121 78 148 17
Carnegie Mellon University - USA 121 121 100 121 121 109 77 121 18
University of Illinois at Urbana-Champaign - USA 79 79 57 79 79 60 53 79 8
Princeton University - USA 68 68 53 68 68 57 40 68 19
Cornell University - USA 65 65 55 65 65 54 37 65 9
University of Washington - USA 61 61 52 61 61 49 40 61 4
Georgia Institute of Technology - USA 53 53 42 53 53 50 42 53 9
Harvard University - USA 49 49 37 49 49 41 32 49 11
University of Texas - Austin - USA 48 48 41 48 48 37 31 48 8
University of Wisconsin - Madison - USA 43 43 31 43 43 29 25 43 6
University of Pennsylvania - USA 41 41 31 41 41 34 30 41 8
University of California - Los Angeles - USA 34 34 29 34 34 29 28 34 8
University of Michigan - USA 34 34 25 34 34 28 22 34 5
University of Maryland - College Park - USA 33 33 28 33 33 25 23 33 2
California Institute of Technology - USA 32 32 20 32 32 27 23 32 11
Columbia University - USA 30 30 28 30 30 23 21 30 6
Purdue University - USA 26 26 15 26 26 20 20 26 3
University of North Carolina - Chapel Hill - USA 25 25 18 25 25 19 15 25 3
Yale University - USA 25 25 18 25 25 21 17 25 1
University of Southern California - USA 25 25 22 25 25 20 19 25 6
University of Massachusetts - Amherst - USA 24 24 17 24 24 20 18 24 5
University of Toronto - Canada 24 24 20 24 24 19 18 24 8
University of California - San Diego - USA 22 22 19 22 22 16 15 22 4
Brown University - USA 22 22 17 22 22 20 19 22 5
State University of New York - Stony Brook - USA 20 20 18 20 20 14 15 20 1
Pennsylvania State University - USA 20 20 15 20 20 15 13 20 3
New York University - USA 20 20 18 20 20 16 15 20 4
University of Minnesota - Twin Cities - USA 19 19 14 19 19 12 11 19 2
Rice University - USA 19 19 15 19 19 15 12 19 1
University of Utah - USA 17 17 15 17 17 16 11 17 4
University of Rochester - USA 16 16 13 16 16 14 13 16 5
Ohio State University - USA 16 16 12 16 16 13 12 16 2
University of Virginia - USA 15 15 14 15 15 14 15 15 0
University of California - Irvine - USA 14 14 7 14 14 11 9 14 3
University of Chicago - USA 14 14 13 14 14 9 8 14 3
Hebrew University of Jerusalem - Israel 13 13 10 13 13 8 7 13 5
University of Colorado Boulder - USA 12 12 8 12 12 11 7 12 1
Duke University - USA 12 12 10 12 12 10 7 12 6
Johns Hopkins University - USA 12 12 7 12 12 11 10 12 0
Rutgers - State University of New Jersey - New Brunswick - USA 10 10 9 10 10 8 5 10 1
University of California - Santa Barbara - USA 9 9 8 9 9 9 6 9 1
North Carolina State University - USA 8 8 3 8 8 7 6 8 0
University of Pittsburgh - USA 8 8 6 8 8 6 6 8 1
Weizmann Institute of Science - Israel 8 8 6 8 8 3 4 8 2
University of Edinburgh - United Kingdom 7 7 6 7 7 6 4 7 1
University of Arizona - USA 7 7 5 7 7 5 5 7 1
Northwestern University - USA 7 7 7 7 7 6 6 7 0
Washington University - St. Louis - USA 7 7 7 7 7 7 6 7 0
University of Oxford - United Kingdom 6 6 3 6 6 4 4 6 0
Tel Aviv University - Israel 6 6 6 6 6 4 3 6 2
Swiss Federal Institute of Technology (ETH) - Switzerland 6 6 6 6 6 1 3 6 1
University of Waterloo - Ontario - Canada 6 6 6 6 6 6 6 6 1
Syracuse University - USA 6 6 3 6 6 4 5 6 0
University of California - Davis - USA 5 5 5 5 5 5 4 5 0
University of California - Santa Cruz - USA 5 5 3 5 5 4 4 5 0
University of Cambridge - United Kingdom 5 5 4 5 5 4 4 5 2
University of Iowa - USA 5 5 4 5 5 4 4 5 1
University of British Columbia - Canada 4 4 4 4 4 4 4 4 1
... ... ... ... ... ... ... ... ...

207 rows × 9 columns

MIT is the largest university, but it seems only a number of university graduates could secure academic jobs in the market. Top 4 universities are responsible for 33% faculty in the universities and the distribution has a long tail.

In [22]:
 
Out[22]:
<ggplot: (278592569)>

MIT and Berkeley is consistently high over time and Stanford are relatively new, but still very high comparing to other universities. CMU exceeds sometimes Berkeley in the recent years in the number of graduates.

In [24]:
 
Out[24]:
Name University Rank Subfield Bachelors Masters PostDoc doctorate_count Year Doctorate
JoinYear Doctorate
1960 University of Texas - Austin - USA 1 1 1 1 1 0 0 1 1960 University of Texas - Austin - USA
1965 University of Pennsylvania - USA 1 1 1 1 1 1 0 1 1965 University of Pennsylvania - USA
1966 University of California - Los Angeles - USA 1 1 1 1 1 1 0 1 1966 University of California - Los Angeles - USA
1968 University of Wisconsin - Madison - USA 1 1 1 1 1 1 0 1 1968 University of Wisconsin - Madison - USA
1970 University of Michigan - USA 1 1 1 1 1 0 1 1 1970 University of Michigan - USA
1971 University of California - Los Angeles - USA 1 1 1 1 1 1 0 1 1971 University of California - Los Angeles - USA
1973 Columbia University - USA 1 1 1 1 1 1 0 1 1973 Columbia University - USA
1975 University of California - Los Angeles - USA 1 1 1 1 0 0 1 1 1975 University of California - Los Angeles - USA
University of Michigan - USA 1 1 1 1 1 1 0 1 1975 University of Michigan - USA
1976 University of California - Los Angeles - USA 1 1 1 1 1 1 0 1 1976 University of California - Los Angeles - USA
University of Texas - Austin - USA 2 2 2 2 2 1 0 2 1976 University of Texas - Austin - USA
University of Wisconsin - Madison - USA 1 1 1 1 0 0 0 1 1976 University of Wisconsin - Madison - USA
1977 University of California - Los Angeles - USA 2 2 2 2 2 2 0 2 1977 University of California - Los Angeles - USA
1978 University of California - Los Angeles - USA 1 1 1 1 1 0 1 1 1978 University of California - Los Angeles - USA
University of Maryland - College Park - USA 1 1 1 1 1 0 0 1 1978 University of Maryland - College Park - USA
University of Michigan - USA 1 1 1 1 1 1 0 1 1978 University of Michigan - USA
1979 Columbia University - USA 1 1 1 1 1 0 0 1 1979 Columbia University - USA
University of California - Los Angeles - USA 1 1 1 1 0 0 0 1 1979 University of California - Los Angeles - USA
1980 University of Michigan - USA 1 1 1 1 1 1 0 1 1980 University of Michigan - USA
1981 University of Maryland - College Park - USA 1 1 1 1 0 1 0 1 1981 University of Maryland - College Park - USA
1982 Columbia University - USA 1 1 1 1 1 1 0 1 1982 Columbia University - USA
Purdue University - USA 1 1 1 1 1 1 0 1 1982 Purdue University - USA
University of Maryland - College Park - USA 3 3 3 3 2 3 0 3 1982 University of Maryland - College Park - USA
University of Pennsylvania - USA 1 1 1 1 1 1 0 1 1982 University of Pennsylvania - USA
University of Texas - Austin - USA 1 1 1 1 0 0 0 1 1982 University of Texas - Austin - USA
1983 California Institute of Technology - USA 1 1 1 1 1 1 0 1 1983 California Institute of Technology - USA
Purdue University - USA 1 1 1 1 1 1 0 1 1983 Purdue University - USA
University of Wisconsin - Madison - USA 1 1 1 1 1 1 0 1 1983 University of Wisconsin - Madison - USA
1984 Columbia University - USA 1 1 1 1 1 1 0 1 1984 Columbia University - USA
Purdue University - USA 3 3 3 3 3 2 0 3 1984 Purdue University - USA
University of Maryland - College Park - USA 1 1 1 1 1 1 0 1 1984 University of Maryland - College Park - USA
University of Michigan - USA 1 1 1 1 1 0 1 1 1984 University of Michigan - USA
University of North Carolina - Chapel Hill - USA 1 1 1 1 1 1 0 1 1984 University of North Carolina - Chapel Hill - USA
University of Wisconsin - Madison - USA 1 1 1 1 1 1 0 1 1984 University of Wisconsin - Madison - USA
1985 University of Wisconsin - Madison - USA 1 1 1 1 1 0 0 1 1985 University of Wisconsin - Madison - USA
1986 Columbia University - USA 1 1 1 1 1 1 0 1 1986 Columbia University - USA
1987 Purdue University - USA 1 1 1 1 1 1 0 1 1987 Purdue University - USA
University of Maryland - College Park - USA 2 2 2 2 2 1 0 2 1987 University of Maryland - College Park - USA
University of North Carolina - Chapel Hill - USA 2 2 2 2 2 2 0 2 1987 University of North Carolina - Chapel Hill - USA
University of Texas - Austin - USA 2 2 2 2 1 1 0 2 1987 University of Texas - Austin - USA
1988 California Institute of Technology - USA 1 1 1 1 1 1 0 1 1988 California Institute of Technology - USA
University of Michigan - USA 1 1 1 1 1 1 0 1 1988 University of Michigan - USA
1989 University of Maryland - College Park - USA 1 1 1 1 1 1 0 1 1989 University of Maryland - College Park - USA
University of Michigan - USA 1 1 1 1 1 1 0 1 1989 University of Michigan - USA
University of Wisconsin - Madison - USA 1 1 1 1 1 1 0 1 1989 University of Wisconsin - Madison - USA
1990 Columbia University - USA 1 1 1 1 1 0 1 1 1990 Columbia University - USA
University of California - Los Angeles - USA 1 1 1 1 1 1 0 1 1990 University of California - Los Angeles - USA
1991 University of California - Los Angeles - USA 1 1 1 1 1 1 0 1 1991 University of California - Los Angeles - USA
University of Maryland - College Park - USA 1 1 1 1 0 1 0 1 1991 University of Maryland - College Park - USA
University of Michigan - USA 1 1 1 1 1 0 0 1 1991 University of Michigan - USA
University of North Carolina - Chapel Hill - USA 1 1 1 1 1 1 0 1 1991 University of North Carolina - Chapel Hill - USA
1992 University of California - Los Angeles - USA 1 1 1 1 1 1 0 1 1992 University of California - Los Angeles - USA
University of Michigan - USA 1 1 1 1 1 1 0 1 1992 University of Michigan - USA
1993 University of California - Los Angeles - USA 1 1 1 1 1 1 0 1 1993 University of California - Los Angeles - USA
University of Maryland - College Park - USA 1 1 1 1 1 1 0 1 1993 University of Maryland - College Park - USA
University of Texas - Austin - USA 2 2 2 2 2 1 0 2 1993 University of Texas - Austin - USA
1994 Purdue University - USA 1 1 1 1 1 1 0 1 1994 Purdue University - USA
University of Texas - Austin - USA 1 1 1 1 1 1 0 1 1994 University of Texas - Austin - USA
University of Wisconsin - Madison - USA 1 1 1 1 1 1 0 1 1994 University of Wisconsin - Madison - USA
1995 Columbia University - USA 1 1 1 1 1 1 0 1 1995 Columbia University - USA
... ... ... ... ... ... ... ... ... ...

172 rows × 10 columns

In [25]:
 
Out[25]:
<ggplot: (278593601)>
In [26]:
 
Out[26]:
Name University JoinYear Rank Subfield Bachelors Masters Doctorate PostDoc
PostDoc
Massachusetts Institute of Technology - USA 36 36 25 36 36 29 26 36 36
University of California - Berkeley - USA 24 24 24 24 24 20 18 24 24
Princeton University - USA 24 24 20 24 24 18 14 24 24
Stanford University - USA 22 22 20 22 22 18 9 22 22
Carnegie Mellon University - USA 16 16 15 16 16 16 9 16 16
Microsoft Research - USA 16 16 16 16 16 14 6 16 16
Harvard University - USA 16 16 14 16 16 14 10 16 16
University of Washington - USA 13 13 11 13 13 9 9 13 13
University of Pennsylvania - USA 10 10 9 10 10 9 7 10 10
IBM Thomas J. Watson Research Center - USA 7 7 7 7 7 6 6 7 7
University of Illinois at Urbana-Champaign - USA 7 7 6 7 7 6 6 7 7
University of Southern California - USA 7 7 6 7 7 6 5 7 7
Cornell University - USA 6 6 6 6 6 5 3 6 6
University of California - San Diego - USA 6 6 6 6 6 6 4 6 6
California Institute of Technology - USA 6 6 6 6 6 6 4 6 6
Bell Laboratories - USA 5 5 5 5 5 4 3 5 5
Georgia Institute of Technology - USA 5 5 5 5 5 5 5 5 5
University of Colorado Boulder - USA 4 4 4 4 4 4 2 4 4
University of Massachusetts - Amherst - USA 4 4 3 4 4 3 3 4 4
University of Maryland - College Park - USA 4 4 4 4 4 4 3 4 4
Columbia University - USA 4 4 4 4 4 2 2 4 4
University of California - Los Angeles - USA 4 4 4 4 4 3 2 4 4
University of Utah - USA 4 4 4 4 4 4 3 4 4
Duke University - USA 3 3 2 3 3 2 3 3 3
University of California - Santa Barbara - USA 3 3 3 3 3 1 2 3 3
Weizmann Institute of Science - Israel 3 3 1 3 3 3 1 3 3
University of Chicago - USA 3 3 3 3 3 2 1 3 3
IBM Research - Zurich - Switzerland 3 3 2 3 3 3 2 3 3
Johns Hopkins University - USA 3 3 2 3 3 3 2 3 3
University of Oxford - United Kingdom 3 3 2 3 3 3 2 3 3
Hebrew University of Jerusalem - Israel 2 2 2 2 2 0 0 2 2
Technion-Israel Institute of Technology - Israel 2 2 1 2 2 2 1 2 2
Harvard Medical School 2 2 2 2 2 2 1 2 2
Lawrence Livermore National Laboratory - USA 2 2 2 2 2 2 1 2 2
Google - USA 2 2 2 2 2 2 2 2 2
Ohio State University - USA 2 2 2 2 2 2 1 2 2
Los Alamos National Laboratory - USA 2 2 1 2 2 2 0 2 2
University College London - United Kingdom 2 2 2 2 2 2 1 2 2
Yale University - USA 2 2 2 2 2 2 1 2 2
University of California - San Francisco - USA 2 2 1 2 2 2 1 2 2
University of Edinburgh - United Kingdom 2 2 2 2 2 1 1 2 2
University of Texas - Austin - USA 2 2 2 2 2 2 2 2 2
University of Toronto - Canada 2 2 2 2 2 2 2 2 2
Brown University - USA 2 2 2 2 2 2 2 2 2
University of Virginia - USA 2 2 2 2 2 2 1 2 2
University of Grenoble - France 2 2 2 2 2 2 1 2 2
University of Cambridge - United Kingdom 2 2 1 2 2 2 1 2 2
University of California - Santa Cruz - USA 2 2 2 2 2 1 2 2 2
Kyoto University - Japan 1 1 1 1 1 0 0 1 1
Lawrence Berkeley National Laboratory - USA 1 1 1 1 1 1 1 1 1
INRIA - Rocquencourt - France 1 1 1 1 1 1 0 1 1
Broad Institute of MIT & Harvard University - USA - USA 1 1 1 1 1 1 1 1 1
Iowa State University - USA 1 1 1 1 1 1 0 1 1
Bell Laboratories & Microsoft Research - USA 1 1 1 1 1 0 0 1 1
International Computer Science Institute - USA 1 1 1 1 1 1 1 1 1
Intel - USA 1 1 1 1 1 1 1 1 1
Institute of Science and Technology - Austria 1 1 1 1 1 0 1 1 1
Institute for Advanced Study - USA 1 1 1 1 1 1 1 1 1
Facebook - USA 1 1 1 1 1 1 1 1 1
ICSI (International Computer Science Institute - USA) Center for Internet Research - USA 1 1 1 1 1 1 1 1 1
... ... ... ... ... ... ... ... ...

119 rows × 9 columns

MIT also leads in this one as well. For postdocs, some faculty also choose industry(Microsoft, IBM, Bell Labs).

In [30]:
 
Out[30]:
<ggplot: (278072573)>

Berkeley is the best in recent years when we look at the numbers over time.

What is next?

Do not forget to check out the interactive visualization

comments powered by Disqus