88
data science & the design process fickleness and how we solve it Bo Peng

Data Science and Design: Fickleness and How We Solve It

  • Upload
    bop

  • View
    286

  • Download
    0

Embed Size (px)

DESCRIPTION

This is a talk I gave on Jan 31, 2014 as part of Harvard's Institute for Applied Computational Science (IACS) seminar series. Abstract: When solving problems, data scientists often encounter added layers of complexity when the problems to be solved are not well defined, and their solutions unclear. In these cases, standard, more straightforward approaches fall short, as they are not amenable to vague problems, and are thus not guaranteed to reliably produce useful results. At Datascope Analytics, we adopt methodologies from the design community and use a "continuous feedback loop" to iteratively improve dashboards, algorithms, and data sources to ensure that the resulting tool will be useful and well received. During this talk, I will illustrate our approach by sharing a detailed example from one of our projects.

Citation preview

Page 1: Data Science and Design: Fickleness and How We Solve It

data science & the design process fickleness and how we solve it

Bo Peng

Page 2: Data Science and Design: Fickleness and How We Solve It
Page 3: Data Science and Design: Fickleness and How We Solve It
Page 4: Data Science and Design: Fickleness and How We Solve It
Page 5: Data Science and Design: Fickleness and How We Solve It

design is a process.

Page 6: Data Science and Design: Fickleness and How We Solve It
Page 7: Data Science and Design: Fickleness and How We Solve It
Page 8: Data Science and Design: Fickleness and How We Solve It
Page 9: Data Science and Design: Fickleness and How We Solve It
Page 10: Data Science and Design: Fickleness and How We Solve It

LINEAR!

Page 11: Data Science and Design: Fickleness and How We Solve It

the design process - an overview

Page 12: Data Science and Design: Fickleness and How We Solve It

our design process - an overview

Page 13: Data Science and Design: Fickleness and How We Solve It

our design process - an overview

Page 14: Data Science and Design: Fickleness and How We Solve It

our design process - an overview

Page 15: Data Science and Design: Fickleness and How We Solve It

iterate, iterate, iterate

Page 16: Data Science and Design: Fickleness and How We Solve It

and now, the example.

Page 17: Data Science and Design: Fickleness and How We Solve It

informal beginnings

Page 18: Data Science and Design: Fickleness and How We Solve It

the client

•  fortune 50 international company •  major restructuring of 20,000+ personnel

Page 19: Data Science and Design: Fickleness and How We Solve It

initial brainstorm

Page 20: Data Science and Design: Fickleness and How We Solve It

initial brainstorm

Page 21: Data Science and Design: Fickleness and How We Solve It

the data

•  every person has attributes: function, location, division, etc

Page 22: Data Science and Design: Fickleness and How We Solve It

the data •  everyone is encouraged to write about their

work •  when people write, they are encouraged to

write together •  most people took a survey

Page 23: Data Science and Design: Fickleness and How We Solve It

example data

Page 24: Data Science and Design: Fickleness and How We Solve It

On Chemical X and its Y-related effects on diaper absorption

Author: Charlie Brown Date: 31 Jan 2012

Lorem Ipsum is a dummy text used when typesetting or marking up documents. It has a long history starting from the 1500s and is still used in digital millennium for typesetting electronic documents, page designs, etc. In itself, the original text of Lorem Ipsum might have been taken from an ancient Latin book that was written about 50 BC. Nevertheless, Lorem Ipsum's words have been changed so they don't read as a proper text.  Naturally, page designs that are made for text documents must contain some text rather than placeholder dots or something else. However, should they contain a proper English words and sentences almost every reader will deliberately try to interpret it eventually missing the design itself.  However, a placeholder text must have a natural distribution of letters and punctuation or otherwise the markup will look strange and unnatural. That's what Lorem Ipsum helps to achieve.

I would like to thank Peppermint Patty for her support on studying Chemical X, as well the infinite wisdom of

Linus van Pelt and his willingness to use his blanket in my experiments.

Page 25: Data Science and Design: Fickleness and How We Solve It

On Chemical X and its Y-related effects on diaper absorption

Author: Charlie Brown Date: 31 Jan 2012

Lorem Ipsum is a dummy text used when typesetting or marking up documents. It has a long history starting from the 1500s and is still used in digital millennium for typesetting electronic documents, page designs, etc. In itself, the original text of Lorem Ipsum might have been taken from an ancient Latin book that was written about 50 BC. Nevertheless, Lorem Ipsum's words have been changed so they don't read as a proper text.  Naturally, page designs that are made for text documents must contain some text rather than placeholder dots or something else. However, should they contain a proper English words and sentences almost every reader will deliberately try to interpret it eventually missing the design itself.  However, a placeholder text must have a natural distribution of letters and punctuation or otherwise the markup will look strange and unnatural. That's what Lorem Ipsum helps to achieve.

I would like to thank Peppermint Patty for her support on studying Chemical X, as well the infinite wisdom of

Linus van Pelt and his willingness to use his blanket in my experiments.

Page 26: Data Science and Design: Fickleness and How We Solve It

project goals - v. 1

reveal areas of expertise

Page 27: Data Science and Design: Fickleness and How We Solve It

project goals - v. 1

evaluate connectivity of subject matter experts

Page 28: Data Science and Design: Fickleness and How We Solve It

project goals - v. 1 management directed subject matter connections

Page 29: Data Science and Design: Fickleness and How We Solve It

the idea space

Page 30: Data Science and Design: Fickleness and How We Solve It

the idea space

three project goals

Page 31: Data Science and Design: Fickleness and How We Solve It

the idea space

project goal #3

Page 32: Data Science and Design: Fickleness and How We Solve It

consider the end user original use case: to help Kathie connect people

Page 33: Data Science and Design: Fickleness and How We Solve It

brainstorm

Page 34: Data Science and Design: Fickleness and How We Solve It
Page 35: Data Science and Design: Fickleness and How We Solve It

project goal #3

Page 36: Data Science and Design: Fickleness and How We Solve It

diverging ideas

project goal #3

Page 37: Data Science and Design: Fickleness and How We Solve It

then, prototype

Page 38: Data Science and Design: Fickleness and How We Solve It
Page 39: Data Science and Design: Fickleness and How We Solve It

research & testing

Page 40: Data Science and Design: Fickleness and How We Solve It

vs.

Page 41: Data Science and Design: Fickleness and How We Solve It
Page 42: Data Science and Design: Fickleness and How We Solve It

Subject matter experts

¤  Extrusion   Nanotubes   Machining   Web handling   Manufacturing

Page 43: Data Science and Design: Fickleness and How We Solve It
Page 44: Data Science and Design: Fickleness and How We Solve It

vs.

Page 45: Data Science and Design: Fickleness and How We Solve It
Page 46: Data Science and Design: Fickleness and How We Solve It

connected by ý survey ý co-authors ☐ co-expertise

Page 47: Data Science and Design: Fickleness and How We Solve It

expertise percentile machining

manufacturing

web handling

extrusion

nanotubes

Page 48: Data Science and Design: Fickleness and How We Solve It

vs.

Page 49: Data Science and Design: Fickleness and How We Solve It

vs.

Page 50: Data Science and Design: Fickleness and How We Solve It

extrusion nanotubes machining web handing manufacturing

2011 2012

Page 51: Data Science and Design: Fickleness and How We Solve It

vs.

Page 52: Data Science and Design: Fickleness and How We Solve It

diverging

converging ideas

Page 53: Data Science and Design: Fickleness and How We Solve It

The design process involves frequent iterations throughout.

Page 54: Data Science and Design: Fickleness and How We Solve It

Cambridge, MA, USA

Page 55: Data Science and Design: Fickleness and How We Solve It

Cambridge, MA, USA

Page 56: Data Science and Design: Fickleness and How We Solve It

Where’s Waldo?

[pic of static expert finder and stained glass]

Page 57: Data Science and Design: Fickleness and How We Solve It

diverging

converging

iteration

done

Page 58: Data Science and Design: Fickleness and How We Solve It

The design process allows for the development and changes

to client requests

Page 59: Data Science and Design: Fickleness and How We Solve It

back to our design process

Page 60: Data Science and Design: Fickleness and How We Solve It

reveal areas of expertise. evaluate connectivity within experts.

Page 61: Data Science and Design: Fickleness and How We Solve It
Page 62: Data Science and Design: Fickleness and How We Solve It

reveal areas of expertise. evaluate connectivity within experts.

diverging

diverging

Page 63: Data Science and Design: Fickleness and How We Solve It

observing areas of expertise from a macroscopic level.

converging

Page 64: Data Science and Design: Fickleness and How We Solve It

iteration

Page 65: Data Science and Design: Fickleness and How We Solve It
Page 66: Data Science and Design: Fickleness and How We Solve It

let’s compare countries.

Page 67: Data Science and Design: Fickleness and How We Solve It
Page 68: Data Science and Design: Fickleness and How We Solve It

+ 1!

Page 69: Data Science and Design: Fickleness and How We Solve It
Page 70: Data Science and Design: Fickleness and How We Solve It
Page 71: Data Science and Design: Fickleness and How We Solve It

10 5 5 20

8 25 2 5

12 3 30 10

1 20 25 50

Page 72: Data Science and Design: Fickleness and How We Solve It
Page 73: Data Science and Design: Fickleness and How We Solve It
Page 74: Data Science and Design: Fickleness and How We Solve It
Page 75: Data Science and Design: Fickleness and How We Solve It

Dashboard #2: Stained Glass

Page 76: Data Science and Design: Fickleness and How We Solve It
Page 77: Data Science and Design: Fickleness and How We Solve It

iteration

iteration

Page 78: Data Science and Design: Fickleness and How We Solve It

back to our design process

Page 79: Data Science and Design: Fickleness and How We Solve It

Film Film

Page 80: Data Science and Design: Fickleness and How We Solve It
Page 81: Data Science and Design: Fickleness and How We Solve It

iteration

iteration

iteration

Page 82: Data Science and Design: Fickleness and How We Solve It
Page 83: Data Science and Design: Fickleness and How We Solve It

Bing Crosby

Page 84: Data Science and Design: Fickleness and How We Solve It
Page 85: Data Science and Design: Fickleness and How We Solve It

billy

Page 86: Data Science and Design: Fickleness and How We Solve It
Page 87: Data Science and Design: Fickleness and How We Solve It
Page 88: Data Science and Design: Fickleness and How We Solve It

the idea space

converge

converge

diverge

diverge

iterate

iterate