Five Mindsets to Succeed as a Data Scientist IRL Pallav Agrawal, Director, Data Science Levi Strauss & Co.
I WANT TO KNOW YOU
What do Data Scientists Create?
Create Concise Generalizations
“90 percent of the data in the world today has been created in the last two years alone” - IBM Marketing Study
https://twitter.com/Johnny_Uzan/status/1031742658048352257
Data Science source
Data Science is a lot like Observational Astronomy
-- Occam’s razor
Develop a really good BS Filter
source
https://www.facebook.com/dan.ariely/posts/904383595868
source source Companies with ‘.ai’ domains raise 3.5x more money
https://twitter.com/xaprb/status/930674776317849600
Data Science source
“To ascend this pyramid fast…” source
“First build this pyramid you must…” source
Executives AI Startups The Dunning-Kruger Effect
‘AI’ as a Service powers the Flywheel….
“Bullshit involves language, statistical figures, data graphics, and other forms of presentation intended to persuade by impressing and overwhelming a reader or listener, with a blatant disregard for truth and logical coherence.” https://callingbullshit.org/
Learn to Think Like a Lawyer
What is a Sandwich?
Is this a Sandwich?
Is this a Sandwich?
Is this a Sandwich?
Is this a Sandwich?
Is this a Sandwich?
Are any of these a Sandwich?
“SANDWICHES” AS USED IN ARTICLES 47 AND 48 OF TITLE 12, C.R.S. ARE DEFINED AS SINGLE SERVING ITEMS SUCH AS HAMBURGERS, HOT DOGS, FROZEN PIZZAS, BURRITOS, CHICKEN WINGS, ETC.” - Colorado Department of Revenue, Liquor Enforcement Division source
“I know it, when I see it”
What about Fraud Detection?
Or Predicting CLTV?
It often helps to frame a problem statement like a Legal Contract
Kozyrkov’s Razor “When decision-makers don’t realize that thinking deeply is their job, remind them.” source
Utilize UI/UX to Improve CX
“FB Messenger Bot was unable to understand 70% of customer queries” - Motley Fool
“41% of consumers say they stopped shopping with a company because of “poor personalization.” -Accenture
Design for Trust
Encourage User Feedback
Wizard of Oz Testing Start with an MVP Show, Don’t Tell
“You have to start with the customer experience and work backwards to the technology” Steve Jobs
Three (timeless) books to get you started
Data Science is a Team Sport
Types of Data Science Practiced Today: Marketing/PR Centric: Where the hype around ‘AI’ is used to improve perception of an aging company among Wall Street Analysts
Types of Data Science Practiced Today: Marketing/PR Research Centric: Centric: Where the hype Where the around ‘AI’ is discovery of used to new algorithms improve is central to perception of maintain the an aging company’s company among competitive Wall Street advantage Analysts
Types of Data Science Practiced Today: Marketing/PR Research Recruiting Centric: Centric: Centric: Where the hype Where the Where the only around ‘AI’ is discovery of way the used to new algorithms company can improve is central to attract the perception of maintain the best minds is an aging company’s through “look company among competitive at all this Wall Street advantage cool AI we are Analysts doing!”
Types of Data Science Practiced Today: Marketing/PR Research Recruiting Ego-Centric: Centric: Centric: Centric: Where the CEO wants to Where the hype Where the Where the only appear a around ‘AI’ is discovery of way the visionary at used to new algorithms company can Techcrunch improve is central to attract the Disrupt by perception of maintain the best minds is unloading a an aging company’s through “look bunch of company among competitive at all this Singularity Wall Street advantage cool AI we are references Analysts doing!”
Types of Data Science Practiced Today: Marketing/PR Research Recruiting Ego-Centric: Business Centric: Centric: Centric: Where the CEO Centric: wants to Where the hype Where the Where the only Where the most appear a around ‘AI’ is discovery of way the important visionary at used to new algorithms company can business Techcrunch improve is central to attract the decisions are Disrupt by perception of maintain the best minds is made using unloading a an aging company’s through “look insights bunch of company among competitive at all this derived from Singularity Wall Street advantage cool AI we are the scientific references Analysts doing!” use of data
Data Science as a Types of Data Science Practiced Today: Team Sport Marketing/PR Research Recruiting Ego-Centric: Business Centric: Centric: Centric: Where the CEO Centric: wants to Where the hype Where the Where the only Where the most appear a around ‘AI’ is discovery of way the important visionary at used to new algorithms company can business Techcrunch improve is central to attract the decisions are Disrupt by perception of maintain the best minds is made using unloading a an aging company’s through “look insights bunch of company among competitive at all this derived from Singularity Wall Street advantage cool AI we are the scientific references Analysts doing!” use of data
Business-Centric Data Science is a Team Sport UI/UX Technical Product Manager Business Data Scientist Stakeholder Subject Matter DevOps Expert Project Manager Data Engineer Customer Insights Data Analyst Cheerleader
Roadmap Development Customer Business Technology
Roadmap Development I want: Customer ● Free 1-hour delivery on orders ● Beyonce’s look from Coachella ● To experience the feel of the garment virtually Desirability Business Technology
Roadmap Development I want: Customer ● Free 1-hour delivery on orders ● Beyonce’s look from Coachella ● To experience the feel of the garment virtually Desirability Viability Business Technology
Roadmap Development I want: Customer ● Free 1-hour delivery on orders ● Beyonce’s look from Coachella ● To experience the feel of the garment virtually Desirability Viability Feasibility Business Technology
Roadmap Development I want: Customer ● Free 1-hour delivery on orders ● Beyonce’s look from Coachella ● To experience the feel of the garment virtually Desirability Viability Feasibility Business Technology
Ideation Uncovering Automation Opportunities: If I could identify/recognize/interpet _____________________________________________ in __________________________________________, I could _____________________________________. Hint: sources could include anything from images, video, audio, text or a mix of these. http://milkandhoney.ai/
Ideation Uncovering Prediction Opportunities: If I could predict precisely how much/many _____________________________________________ at any given moment, I could _____________________________________________. Hint: identify places where you currently rely on estimates and averages to make decisions. If I could predict the fastest way to _____________________________________________ at any given moment, I could _____________________________________________. Hint: identify processes that require moving something from point A to point B with multiple paths to choose from. http://milkandhoney.ai/
RECAP Before executing, first craft a precise problem statement developed with decision makers using data and subject matter expertise Data Science is the process of Focus on improving the customer extracting concise and actionable experience by making the customer insights from data through journey intuitive and frictionless scientific rigor Practice healthy skepticism towards Learn to communicate and most remarkable claims, and collaborate within cautious optimism towards the interdisciplinary teams to build applicability of positive results products that lie at the intersection of desirability, viability and feasibility
Questions?
Are you doing analytics or statistics? Cargo Cult science? Some Key Principles • use many data sources (the plural of anecdote is not data) • understand how the data were collected (sampling is essential) • weight the data thoughtfully (not all polls are equally good) • use statistical models (not just hacking around in Excel) • understand correlations (e.g., states that trend similarly) • think like a Bayesian, check like a frequentist (reconciliation) • have good communication skills (What does a 60% probability even mean? How can we visualize, validate, and understand the conclusions?)
Recommend
More recommend
