This document discusses cultivating creativity in data work. It argues that data science involves both analysis (Type A) and building (Type B). It emphasizes learning theory, mastering tools/code, and composing narratives to convey insights. Errors are viewed as system failures, not personal faults. Design methods like interviews and prototyping can be applied to data problems through "data design sprints". Empathy and creative delivery are important for success beyond just scientific methods.
2. WOMEN IN ANALYTICS CONFERENCE
ABOUT ME
▸ Data Scientist at Stitch Fix
▸ Formerly a Data Analyst at Etsy
▸ #rcatlady
▸ Co-host of “Not So Standard Deviations”
podcast
28. THEORY
▸ Third person perspective / observational science
▸ Proofs for properties of statistical tests and phenomena
29. THEORY
▸ Third person perspective / observational science
▸ Proofs for properties of statistical tests and phenomena
▸ Empirically explored
30. THEORY
▸ Third person perspective / observational science
▸ Proofs for properties of statistical tests and phenomena
▸ Empirically explored
▸ Provides specific statements that can be used to construct the analysis
narrative.
34. WE MAY FIND THAT THE TWO BOTTLENECKS ARE WHAT YOU
WANT TO DO, AND HOW YOU TELL THE COMPUTER TO DO THAT. A
LOT OF MY EXISTING WORK…HAS BEEN MORE ABOUT HOW TO
MAKE IT EASIER TO EXPRESS WHAT YOU WANT
Hadley Wickham
TEXT
https://statr.me/2013/09/a-conversation-with-hadley-wickham/
38. BLAMELESS POSTMORTEM
▸ Rather than viewing errors as “human error” / person making a mistake, view
them as the system failing a person with good intentions
39. BLAMELESS POSTMORTEM
▸ Rather than viewing errors as “human error” / person making a mistake, view
them as the system failing a person with good intentions
▸ Adapt the system so that it does not fail a person with good intentions
40. COMMON ERRORS (THAT HURT FLUENCY)
▸ You re-run the analysis and get different results
41. COMMON ERRORS (THAT HURT FLUENCY)
▸ You re-run the analysis and get different results
▸ Your data becomes corrupted, but you don’t notice
42. COMMON ERRORS (THAT HURT FLUENCY)
▸ You re-run the analysis and get different results
▸ Your data becomes corrupted, but you don’t notice
▸ Requests and questions from your collaborators get lost
43. COMMON ERRORS (THAT HURT FLUENCY)
▸ You re-run the analysis and get different results
▸ Your data becomes corrupted, but you don’t notice
▸ Requests and questions from your collaborators get lost
▸ …
53. THE A-HA MOMENT
▸ Observable only from the first-person perspective
▸ Third person observers can only rely on accounts
54. THE A-HA MOMENT
▸ Observable only from the first-person perspective
▸ Third person observers can only rely on accounts
▸ People are unreliable about communicating their experiences
62. DESIGN ABILITY IS, IN FACT, ONE OF THE THREE FUNDAMENTAL
DIMENSIONS OF HUMAN INTELLIGENCE. DESIGN, SCIENCE,
AND ART FORM AN ‘AND’ NOT AN ‘OR’ RELATIONSHIP TO
CREATE THE INCREDIBLE HUMAN COGNITIVE ABILITY.”
Nigel Cross, Designerly Ways of Knowing
68. DATA DESIGN SPRINTS
▸ How do we design a study to understand our customers better?
▸ How do we design a plan for developing a new feature for our website?
69. DATA DESIGN SPRINTS
▸ How do we design a study to understand our customers better?
▸ How do we design a plan for developing a new feature for our website?
▸ How do we incorporate new data from users into multiple downstream use-
cases?
70.
71.
72. ▸ Consensus on:
▸ Architecture of problem
▸ First wave of statistical approaches
▸ Next steps (and ownership of these steps)
73. SOME RESOURCES
▸ Designing Your Life
▸ Articulating Design Decisions
▸ Statistics as Principled Argument