Workshop on Data Science and
Decision Making
Dr. Patricia Bagsby & Abhimanyu Gupta
September 12-16, 2022
~clj~}FE~N
!~
• SA INT LOU IS UN IVERSITY
Welcome!
What to expect:
• Discuss the process of how you make decisions
• Adult learning theory
• Executive decision-making models
• Outline some influences in the process of problem solving
• Bias
• Growth Mindset
• Receive problem statement you will work on over the next 4 days
Importance of the Decision Maker
• Decisions are only as good as the decision maker(s)
• Coding influenced by the perspective of the coder
• Racist, classist and sexist outcomes in algorithms
• Google apologized for photo algorithm mislabeling people of color as animals 1
• Amazon's face match Al falsely matched members of US Congress with mugshots; und
those were disproportionately people of color2
• Researchers in US found a program (COMPAS) used to predict criminal reoffence
overclassified women as higher risk compared to actual behavior3
• Grading algorithm was used to generate grades for students in Great Britain during
pandemic - over inflated grades for private schools, under valued grades for public
schools1
• Juries make better decisions (e.g., consider evidence more fully, and evaluate
from multiple perspectives) when jury is diverse!>
SourC('S:
1. http•://www.cb•nrws.com/nrws/i:ooitlc-phoro,i•lnbrlcd•pirs•of••friron••mrrir•M•M•JtOrill••/
2, https://wMv.aclu.or'l,/IICW1/pri,·aC)'•lCChnolo1tY/ ftUIIW>l1'•fot't'·rttOgltllio11-falscly-111•1rbcd-21I
:l, htl(l<:/ /'""'·"'•• nrrhgMo, nol/pnbli<nllon/:l.'121 1:1214_ Tho_....i.r_nlgoritlnn
1. htt1>•://•mw.l.,l.,c.cum/11cw•/ •-duc,,tiu11•537tl7203
:,, httv•://www.u111lri11J,,..,111;,livonity-ul•jury•(XXll-crilia1l·t<>-lair-outw1110¥/1tm1>/
IAndragogy: The Art and Science of Helping Adult Learn I
Self-Concept Readiness Motivation Experience Orientation
W,nt to lt1rn thl nas H, ve brudlh & dtpt h of ngage y earn ng
Ad ults are sel f• relevant to ,,. 1 to-life,. Int rinsically
t-)!per!ence, to draw o n that Is problem •
directed cont••t motlvaled by various and a pply 10 n1w centered and
value drivers learnlna racl col
Goal focused
Independent Have diverse Are practical -
Want just -in-time Self-esteem experience & learning should apply
learn ing knowledge to their lives or job
Like to find their
own way Value meaningful M ay have Want to be Involved
learning experl,.nces Quality of hfe ingrained ideas In planning their
Can make their about things learning
own decisions N~tld I clt:U · why• and
'"what • for l•1rn 1na gaal5
Use problem -solving, Focused on aspects
Want to manage Personal grown &
critical reflection and that are most useful
their own learning M otivation development reasoning skills to them
Source: Adapted from https://instrnctionaldesign.com.au/andragogy-adult-learning-theory/
How Your Brain Decides T l lrNK JNt;,
l'AST ..• SLOW
• Thinking Fast and Slow, :
DAN I EL
• System 1 thinking = your base instincts KAII NEMAN
• System 2 thinking = your rational function
Syst ~m 2 thinking
FOSI O"<I ln$Un<;Uve • Slow. dellberole and
effortful
• Involve$ lhe omygdolo
and bocot gongllo • lnvotv11 11,e P,e•I rontol
c:orlvr
• Usec heurlcoc, (mental
shortcuts llnlc1d wNII unc IOrge ornounte 01
llobttc and routine mental cnrroy. require,
behaviours) c:oncentrolton ond f ocus
Preferroble for • P1efeuoble for complex
•spllt•second" deetslOns IIO$On!ng. Slrolegtc
tNnttng and processing
• Vulnercble to bJO.ses
c:omplu tnformotlon
L.
Source: https://blog.innerdrive.co.uk/business/how-to-think-like-a-leader
~
ml
~e Science of Decision Making
• How do Executive Make
Decisions?
Level Example
•
•
Should we merge with another org?
Should we pursue a new produce line?
l
• Should we downsize?
• Different Decision Models
Tactical • How can we help employees from merged orgs Managers
work together?
• Level of the decision • How should we market a new product line?
• Who should we let go when we downsize?
• Vulnerable to influence
from system 1 thinking Operational • How should I communicate with my new co- Employees
through: workers? throughout the
• What should I tell customers about the new product organization
line?
• Cognitive short cuts • How will I balance my new work demands?
(heuristics) & Bias
• Selflimiting thought
Source: Adap,ed from http://courses.lumenlearning.org/suny-orgbehavi
mm Wcisions Influence by Bias
• We all have bias
DEF INING THE SEEDS MODEL® OF BIAS D
••
• Inherently, neither good
Similarity: •People Ike me ore better than others. •
nor bad - developed as an
adaptive process to help
humans make informed Expedience: • II It feels right, It must be true. •
••
decisions based on prior
knowledge and experience Experience: • My pera!ptiansare aa:urote. •
Distance: •aaser Is better than distant. •
• Roughly 150 biases -
Neuroleadership Institute
narrowed it down to 5
categories (SEEDS) •
~~
Safety: "Bad Is stranger than good. •
Source: https://www.linkedin.com/business/talent/blog/tal ent-acquisition/why-unconscious-bias-training-do es-n
n•..a M,nd,wt
,Jo,! .,, • ·• ,,
Growth Mindset =
Better Decisions
Growth Mindset
• Antidote to self-limiting
thought
• Growth vs. Fixed Mindset
Source: h llps://academicsuccess.ucf.ed u/ sarc/ molivation-persistencc
1.Identify
the problem
-"'
Problem Solving Process
8. Evaluate 2.Establish
the decision decision
criteria
1' \
• Rational Decision-
I
\
\ Making Model
I ",/
• Information on
- \
.
\
- II-
I
I
alternatives can be
gathered AND quantified
• The decision is high level,
high visibility
• You are trying to
•~- •
<
••••
. . . . maximize your outcome
Source: Adapted from http:/ /W>urses.lumenlearning.org/suny-orgbehavior
mml
Wo!em Solving Process
• Creative Decision-Making
• Solutions to the problem are not clear
• New solutions need to be generated
• You have time to immerse yourself in the issue
• Cannot solve a
Step 2
Immersion
• Consciously
think about
• Step 3
Incubation
problem
aside
- • Insight moment
Step 5
Verification &
Application
• Consciously
verify the
problem you do problem • Brain is • Answer becomes feasibility of
not recognize • Gather working apparent, the solution
information non- sometimes when • Implement
consciously you least expect it solution
Source: Adapted from http:/ /iOurses.lumenlearning.org/suny-orgbehavior
mi 'BY'take Aways
• The decision-maker is just as important as the process for
problem-solving
• Decision makers are vulnerable to System 1 thinking
• System 1 thinking = bias prone, self-limiting thought
• Adults are goal oriented and practical, but prone to ingrained ideas
• Using decision-making processes based on the goals increases
quality of output
• Decision-Making models vary based on the context at hand
• Choosing the right decision-making model keeps your brain in System
2 thinking and likely produces better outcomes
•
Problem S'taten1ent
GhtudCu \:\rl1t~1,lt£.,..t11e ie1 sl niggling ~~ri11I~ en111!tJ} i!e allri Uun ..
11
111 mbt! lasl
qunrh:r, one 1n iotur l!nl[ph~_Vt!t:S 9u1t. " ,Ir. St~r~, IIIP ( EO of Gho~t'Cn 1
huts Jurn1\td u cumndnce l u ·dL-chle on u til rufc~gy ro reduce the ollrilion
1
nh! illi mut'~m as, pws.sibl~.
The cnrnmill~ clmuperson~ Ms. Pumpkia r.\US nbie lo nmisge for
employee daln rehiled _ Le;, othitimn &om lbe HR deparlmeuL She
reach~d mJ l t~ you ta help her [denlif;" U1e roo'l cau:s-cs md COllil'l:It
them 111ln dcctSioll5.
lim-. mm you heJp her?
·w rap .U p
• Day .2:. Pnrtid_pants will lea.·rn. bo. to solve the prob'lem from tbe
w.·
perspective of and using Data Vifiunllzatlon.
"l,
•' Day 3: P,rtid,· ·rs wjll l~un how to s_o,Ivc tl;i~ problclil ·from the. .
PJ'~t>ft.ii.ve o~d ustrtg Unsupemsed Madune Leaming Techn1.ques (e.g.,
Clusler AnalysU1),
• Day 5: P~u1icipants w5}1learn bqw _to,solve t~,e -problem _
fro!n time
·perspective of and, usn1g Prnenpti\'I: APaly1H1 le.g., ,Optimt7iilllio·n).
