A Taste of Data Science Michael Clarkson NACS Executive Leadership - - PowerPoint PPT Presentation

A Taste of Data Science

Michael Clarkson NACS Executive Leadership Program at Cornell August 2019

Download these slides at https://tinyurl.com/TasteDataScience

Computer Science Business Statistics Data Science

Diagram inspired by h"p://drewconway.com/zia/2013/3/26/the-data-science-venn-diagram

Michael Clarkson, PhD Senior Lecturer in Computer Science Cornell University

Launched Data Science For All at Cornell

Photo: http://100photos.time.com/photos/andreas-gursky-99-cent

$

Image: https://pixabay.com/photos/data-computer-internet-online-www-2899901/

Photo: https://www.flickr.com/photos/robertnelson/29699674200 https://blog.blueapron.io/forecasting-demand-at-blue-apron-ba62d6af5da2 h"ps://www.forbes.com/sites/stevebanker/2018/01/02/data-science-and-the-meal-kit-subscrip>on-business-model

Photo: https://stories.starbucks.com/stories/2016/starbucks-mobile-app-launches-in-indonesia h"ps://www.forbes.com/sites/bernardmarr/2018/05/28/starbucks-using-big-data-analy@cs-and-ar@ficial-intelligence-to-boost-performance https://www.forbes.com/sites/bernardmarr/2018/04/04/how-mcdonalds-is-getting-ready-for-the-4th-industrial-revolution-using-ai-big-data-and-robotics https://www.wired.com/story/mcdonalds-big-data-dynamic-yield-acquisition/ h"ps://www.reuters.com/ar@cle/us-mcdonalds-mobile-idUSKBN16L2RM

Photo: https://www.uber.com/en-BE/blog/ubereats-antwerp h"ps://www.wired.com/story/how-data-helps-deliver-your-dinner-on-8me-warm h"ps://www.eater.com/2018/10/24/18018334/uber-eats-virtual-restaurants h"ps://venturebeat.com/2018/10/02/uber-eats-and-the-6b-bookings-run-rate-the-ai-success-story-no-one-is-talking-about

Data science is the key to personalized consumer experience and efficient operations at scale

What is data science?

Answering questions from data using computation

Ask Question Obtain Data Understand Data Understand World

Example: Grocery Delivery

What questions would you ask? What data would you obtain?

https://tech.instacart.com/3-million-instacart-orders-open-sourced-d40d29ead6f2

3,000,000

• rders

200,000 customers

2017 Public Dataset

Photo: https://pxhere.com/en/photo/940160

In- ference Prediction Exploration

identifying patterns making guesses quantifying reliability

In- ference Prediction Exploration

identifying patterns making guesses quantifying reliability

DEMO

Data scientists…

• Organize:

collect and clean data

• Discover and communicate:

explore, program, and visualize

• Automate:

separate data files from analyses for repeatability

In- ference Prediction Exploration

identifying patterns making guesses quantifying reliability

DEMO

Recommendations

Program Input Output

Image: https://en.wikipedia.org/wiki/File:Computer.svg Photo: https://www.flickr.com/photos/nihgov/23682213069 Explanation (Prof. Weinberger): http://www.cs.cornell.edu/courses/cs4780/2018fa/lectures/lecturenote01_MLsetup.html

…

✔ ❌ ❌

Data Program Input Output

✔ ❌ ❌ ✔

Machine Learning:

Programs that improve on some task with experience

https://commons.wikimedia.org/wiki/File:Backgammon_lg.jpg https://pixabay.com/vectors/email-mail-spam-message-e-mail-29853/ https://pixabay.com/vectors/car-automobile-tesla-autonomous-2692593/

https://www.freepik.com/free-photo/3d-render-male-head-showing-brain_1111538.htm

Artificial Intelligence Machine Learning

Beyond bananas…

Photo: https://pixabay.com/illustrations/bananas-fruit-yellow-plant-food-3735673/

Nearest neighbors algorithm

To make a recommendation for bananas to Alice:

• Find the 3 “most similar” customers to Alice
• Have them majority vote on whether they would

recommend bananas

• Use that decision

Image: https://alliance.seas.upenn.edu/~cis520/dynamic/2016/wiki/index.php?n=Lectures.LocalLearning

Data scientists…

• Train computers:

use data to “teach” computer how to do task

• Predict:

given a never-before-seen input, find right output

In- ference Prediction Exploration

identifying patterns making guesses quantifying reliability

DEMO

Data scientists…

• Are skeptical:

always ask, “could it be just random chance?”

• Explain uncertainty:

provide answer + estimate of “how right” answer is

In- ference Prediction Exploration

identifying patterns making guesses quantifying reliability

Privacy

Photo: https://pixabay.com/illustrations/mobile-security-privacy-protected-3469818/

Photo: http://faqhow.com/other/any/how-target-predicted-a-girls-pregnancy-before-her https://www.forbes.com/sites/kashmirhill/2012/02/16/how-target-figured-out-a-teen-girl-was-pregnant-before-her-father-did

Photo: h'ps://www.wired.com/2010/03/ne8lix-cancels-contest

Privacy concerns

• Governments/businesses and individuals are

sometimes at odds over how identity is used

• Intrinsic privacy: the individual's right to be left alone
• Informational privacy: the individual's right to

determine for itself when, how, and to what extent information about it is communicated to others

Ethical data scientists…

• Seek consent
• Select minimal identity
• Limit storage
• Avoid linking

http://www.cs.cornell.edu/fbs/publications/chptr.AuthPeople.pdf

Computer Science Business Statistics Data Science

Diagram inspired by http://drewconway.com/zia/2013/3/26/the-data-science-venn-diagram

Photo: h'ps://news.cornell.edu/stories/2019/05/cornell-celebrate-151st-commencement

Acknowledgments

CS 1380 Data Science For All Cornell University Michael Clarkson and Madeleine Udell Data 8 Foundations of Data Science University of California, Berkeley Ani Adhikari and John DeNero

A Taste of Data Science

Michael Clarkson NACS Executive Leadership Program at Cornell August 2019

Download these slides at https://tinyurl.com/TasteDataScience