A Shiny New World of Programming and Visualization

Michael Man Abstract

♦ Mega-trend and break-through in technologies demand and enable change in how we do statistical computing. The future of computing will need to be strong on 3 key areas: high-performance computing, interactive visualization, rapid programming. The ideal is TEDDIE (totally easy to develop, deploy, interact, and explore) to meet the current and future challenge. I will use an example of TEDDIE in action to illustrate what the Shiny new world is like. In conclusion, it is an exciting time for us to embrace the future of computing to power innovation.

5/23/2016 Company Confidential © 2015 Eli Lilly and Company 2 Wind tunnel

5/23/2016 Company Confidential © 2015 Eli Lilly and Company 3 AlphaGo

♦ The first computer program to beat one of the best Go players in March 2016. ♦ Why? • Distributed computing – 1920 CPUs, 280 GPU • Artificial intelligence – AlphaGo's algorithm uses a combination of neural networks, machine learning and Monte Carlo tree search techniques, combined with extensive training, both from human and computer play. [wikipedia]

hardware + software => A Shiny new world

5/23/2016 Company Confidential © 2015 Eli Lilly and Company 4 Mega-trend & technologies demand and enable change

PC, SAS good

Big data Better Cloud computing Faster More AI

“Big data and analytics are changing how we think and operate. Because of the ever-growing accumulation of electronic data … there is

S.5/23/2016 Ahrweiler, http://www.phusewiki.org/docs/Conference%202014%20PD%20Papers/PD08.pdfan increasedCompany need Confidential for ©smart 2015 Eli Lilly analytics...” and Company ~ Pandu Kulkarni, VP of Eli 5Lilly Exciting time …

♦ Key take-away from “Careers in Statistics Evolve and Expand” • A new field “data science” – a fusion of statistics, computer science, and analytics to deal with big data, identify trends, and exploit knowledge. • “There’s a lot of value to be unlocked by understanding that data” – Hal Varian, Google’s chief economist • “So there's much more demand for people who have the quantitative background in handling large [amounts of] health science data” – Xihong Lin, a well-known professor in Statistics at Harvard • "There's so much more receptiveness now to a statistician walking up to somebody like a lawyer and saying, 'Hey, can we apply quantitative tools to study what you're doing?' “ – Cassandra W. Pattanayak, Guthman director of the Quantitative Analysis Institute at Wellesley College • By Puneet Kollipara, http://www.sciencemag.org/careers/2014/03/careers-statistics- evolve-and-expand

5/23/2016 Company Confidential © 2015 Eli Lilly and Company 6 History and future of computing

♦ Hardware: • mainframe -> PC -> parallel/cloud computing ♦ Software: • complex programming languages -> more efficient, more natural, more cross-pollination … ♦ User experience: • static -> interactive visualization • passive -> active participation in development (programming) ♦ 3 key areas in Computing • computer + programming + visualization

iPhone

5/23/2016 Company Confidential © 2015 Eli Lilly and Company 7 http://www.i-programmer.info/news/98-languages/8809- the-evolution-of-programming-languages.html

Object-oriented programming

http://wikibon.org/blog/wp-content/uploads/2012/02/evolution-of-programming.html

SAS was originally written in PL/I, Fortran, and assembly language. In 1985, SAS was rewritten in the C programming was designed by Ross Ihaka and Robert Gentleman in language. This allowed for the SAS' Multivendor Architecture that 1993. R was influenced by Scheme, S, and XLispStat. https://en.wikipedia.org/wiki/SAS_(software) allows the software to run on UNIX, MS-DOS, and Windows. https://cran.r-project.org/doc/html/interface98-paper/paper_1.html https://en.wikipedia.org/wiki/SAS_(software) 5/23/2016 Company Confidential © 2015 Eli Lilly and Company 8 visualization

5/23/2016 Company Confidential © 2015 Eli Lilly and Company 9 Ideal for future computing

♦ Strong in all 3 areas • High-performance computing • Interactive visualization • Rapid development (programming) ♦ TEDDIE Totally Easy to Develop, Deploy, Interact, and Explore

programming visualization

High-performance computing

5/23/2016 Company Confidential © 2015 Eli Lilly and Company 10 Illustration of the Shiny New World: an example of TEDDIE in action ♦ TEDDIE • Turn mundane into magic – FM radio -> records -> cassettes -> CD -> MP3 music player – Phone -> cordless phone -> cell phone – iPod -> iPhone (smart phone) ♦ BEACH as an example

5/23/2016 Company Confidential © 2015 Eli Lilly and Company 11 Goal & motivation for BEACH

♦ To drive innovation in drug development • Identifying subgroups/markers for precision medicine (patient tailoring) • Simulation for clinical studies/programs • Automation of analyses • Real time analyses & dynamic visualization • Identification and elimination of data errors

5/23/2016 Company Confidential © 2015 Eli Lilly and Company 12 What was NOT working

♦ Outdated technology or process ♦ Time-consuming iterations … ♦ Always re-inventing wheel ♦ No time to think & engage

Necessity is the mother of invention. Pain can be the father of Innovation.

5/23/2016 Company Confidential © 2015 Eli Lilly and Company Where is the Panda?

5/23/2016 Company Confidential © 2015 Eli Lilly and Company 14 Where is the Panda?

Familiarity can be the enemy of innovation.

5/23/2016 Company Confidential © 2015 Eli Lilly and Company 15 Epiphany: Shiny -> PARTY

♦ An epiphany is an experience of sudden and striking realization. ♦ Eric Nantz introduced Shiny to Lilly in the summer of 2013. ♦ Can we use it for TFL automation? ♦ It led us to develop PARTY (Pandu’s Automated Reporting Tool @Lilly).

5/23/2016 Company Confidential © 2015 Eli Lilly and Company 16 BEACH

♦ “boil the ocean” => have a BEACH PARTY ♦ Make it fun w/ F.I.R.E.

♦ Fast generation of TFLs ♦ Interactive visualization & analysis ♦ Reproducible with saved code & LOA ♦ Extendible for thousands of analyses

5/23/2016 Company Confidential © 2015 Eli Lilly and Company 17 Who

♦ Chenchen Yu ♦ Tuan Nguyen ♦ Danni Yu ♦ Michael Man ♦ Other contributors: Sameera Duytrac Nathan Eric Wijayawardana Nguyen Enas Nantz Jonathan James Glen Anita Klinginsmith Dickson Christie Morrison

♦ Sponsors: Pandu Kulkarni, Jon Denne BEACH demo

5/23/2016 Company Confidential © 2015 Eli Lilly and Company 19 One more thing …

5/23/2016 Company Confidential © 2015 Eli Lilly and Company 20 A few golden nuggets in R

• A data visualization package by Hadley Wickham • Grammar of Graphics => easy and elegant plots • Tutorial ♦ Shiny ♦ dplyr - tutorial ♦ %>% - tutorial ♦ RTF & RTF2 – for generating RTF tables ♦ Graphical approach for multiple test procedures • tutorial

5/23/2016 Company Confidential © 2015 Eli Lilly and Company 21 Conclusion

♦ Huge benefit for embracing the Shiny New World to power innovation • TEDDIE • Connect w/ open source ecosystems • Access to a huge talent pool • Develop iron men who bring innovation -> standing on shoulders of giants • It is FREE (most of the time)!

5/23/2016 Company Confidential © 2015 Eli Lilly and Company 22