AT&T Labs - Data Driven NYC // March 2014

Post on 18-May-2015

163 Views

Category:

Technology

0 Downloads

Preview:

Click to see full reader

DESCRIPTION

AT&T Labs Head of Data Science Chris Volinsky presented at March's edition of Data Driven NYC. AT&T Labs is the research & development division of AT&T, where scientists and engineers work to understand and advance innovative technologies relevant to networking, communications, and information.

Transcript

Shaping Cities of the Future using Mobile Data

Chris VolinskyAT&T Labs-Research

@statpumpkin

Wednesday, February 5, 14

Wednesday, February 5, 14

Wednesday, February 5, 14

Wednesday, February 5, 14

Mapping potholes in Boston

Wednesday, February 5, 14

courtesy flickr/hilarymason

Wednesday, February 5, 14

Hold for MTS

Interactive visualization done using nanocubes.net

Wednesday, February 5, 14

No Content Ever

Anonymize (always)

Aggregate (when possible)

Reduce granularity

Principle of Least Privilege

Wednesday, February 5, 14

Wednesday, February 5, 14

Wednesday, February 5, 14

Wednesday, February 5, 14

Wednesday, February 5, 14

Wednesday, February 5, 14

Wednesday, February 5, 14

Wednesday, February 5, 14

Wednesday, February 5, 14

Wednesday, February 5, 14

New York

New Jersey

13

6

10

Wednesday, February 5, 14

R1R2

R3

R4

R5

R6

R7R8

R9

R10

R11 R12

R13

Wednesday, February 5, 14

Wednesday, February 5, 14

Call A A->B Call B

C1C1C1C1C2C2C2C2 C1C1C1C2C2C3C3C3

Earth Mover Distance (EMD)

amount of mass moved

distance mass moved

Wednesday, February 5, 14

6024

47

16

101

86

232

101

30

83 110

42

29

33

Wednesday, February 5, 14

0:1

1 5.4%

0:10:1 2 19.8%

0:1

0:1 3 18.4%

0:1

0:1 4 8.1%

0:1

0:1 5 22.2%

0:1

0:1 6 15.9%

0:1

0:1 7 10.2%

M W F S M W F S

68

noon

35

79

midn.

3 Voice SMS

M W F S M W F S

68

noon

35

79

midn.

3

M W F S M W F S

68

noon

35

79

midn.

3

M W F S M W F S

68

noon

35

79

midn.

3

M W F S M W F S

68

noon

35

79

midn.

3

M W F S M W F S

68

noon

35

79

midn.

3

M W F S M W F S

68

noon

35

79

midn.

3

Clustering Users

Wednesday, February 5, 14

Use real data to create synthetic data that has the same statistical properties.

Wednesday, February 5, 14

Wednesday, February 5, 14

Thanks!volinsky@research.att.com

@statpumpkin

Wednesday, February 5, 14

top related