Session 87 PD, R U up on R? Moderator: David L. Snell, ASA, MAAA. Presenters: Melissa Carruthers, FSA, FCIA

Transcription

1 Session 87 PD, R U up on R? Moderator: David L. Snell, ASA, MAAA Presenters: Melissa Carruthers, FSA, FCIA David L. Snell, ASA, MAAA SOA Antitrust Disclaimer SOA Presentation Disclaimer

2 R U UP On R? Melissa Carruthers, FSA, FCIA Senior Consultant, Deloitte October 2016

3 Agenda An evolving landscape 3 The actuarial scientist 9 Introducing R 15 Getting started 22 Demo in R Studio 27 2

4 An evolving landscape

5 In any given minute 639,800 GB of global IP data is transferred 1.3 million videos 2+ million topics searched 47,000 application downloads 4

6 BIG DATA in life insurance 5

7 The power of predictive analytics Using predictive analytics we can determine the likelihood of a person having certain health characteristics and developing future ailments. Low risk of developing gestational diabetes during pregnancy 10% chance of developing breast cancer Average odds of having hay fever < 2% chance of developing Parkinson s disease Tendency toward higher BMI Greater tendency to overeat Likely lactose intoleran t Strong chance of developing severe nearsightedness 2% chance of developing chronic kidney disease Average odds of developing pancreatic cancer Has wet earwax 4% chance of developing melanoma Average odds of getting gout She is pregnant and the baby is likely to weigh 2 ounces less than average at birth 8% chance of developing rheumatoid arthritis Average sensitivity to sweaty odors Average odds of developing ovarian cancer 1% chance of developing age-related macular degeneration Slow caffeine metabolizer (drinking coffee increases chance of heart attack).1% chance of developing type 1 diabetes 5% chance of developing MS Low odds for high blood pressure Average odds of developing glaucoma Average odds of developing uterine fibroids Average odds of developing esophageal cancer 6

8 Changing the way we work Sampling of Applications of Predictive Analytics in Core Operations for Life Insurers Producer Optimization Product Design & Pricing Sales and Marketing New Business & Underwriting Inforce Management Claims and Fraud Producer Target Marketing / Recruitment Lead Generation Identification of Improve quality of leads individuals most likely by identifying those most to become a successful likely to qualify & most producer for a given likely to buy manufacturer Producer Retention Segmenting existing producers and deploying customized tactics to support success and retention Producer-Client Matching Identify behavioral patterns and personality attributes associated with successful, lasting producer-client relationships; deploy tactics to optimize matches Up-Sell Programs Identify existing customers whose need for life insurance has increased, and who remain healthy. Offer increased face amount with limited underwriting Application Triage Identifying certain healthy individuals for which certain medical exams can be waived Underwriting Predicting mortality experience on a seriatim basis, using new data sources to supplement or replace certain traditional medical exams Cross-Sell Programs Identify existing customers who are likely to need and likely to buy a second product an annuity, a P&C product, etc. Deploy customized, targeted offers. Customer Lifetime Value Enable calculation of customized individual CLV; deploy customized proactive tactics for retention, second offers, etc. Post-Level Term Offers Segment population based on current health risk, current life insurance needs, likelihood to buy. Deploy customized, targeted offers Retention Strategy Use customized, individual estimate of lapse likelihood to enable customized proactive and reactive tactics to improve retention effectiveness LTC Claims Management (Active Lives) For each active life, estimate the likelihood of developing certain cognitive or physical impairments, then proactively encourage healthy policyholder behavior LTC Claims to enable Management prevention (Disabled Lives) For each disabled life, estimate the likelihood of transitions between type of impairment (physical vs. cognitive) and associated level of care required (home health care, assisted care facility, nursing home), then proactively encourage healthy policy holder behavior Fraud Detection Identify potential overpayments of claims for LTC or related products 7

9 Ready for the opportunities ahead? 8

10 The actuarial scientist

11 Predictive text gone wild 10

12 Enter actuaries 11

13 Knowing just enough to be dangerous Programmer "I wouldn't even say R is for programmers. It's best suited for people that have data-oriented problems they're trying to solve, regardless of their programming aptitude, - Matt Adams, a data scientist at Code School, which offers online programming education ASOP 41 the actuarial report, the actuary should state the actuarial findings, and identify the methods, procedures, assumptions, and data used by the actuary with sufficient clarity that another actuary qualified in the same practice area could make an objective appraisal of the reasonableness of the actuary s work as presented in the actuarial report. 12

14 The Right People? Purple People Technical & Analytical Business & Communication Testing & validation Data manipulation Data modelling Data analysis Reporting software Technology alignment Macro-perspective Business knowledge Business commentary Soft skills 13

15 The Right People? Purple People Technical & Analytical Business & Communication Testing & validation Data manipulation Data modelling Data analysis Reporting software Actuary Technology alignment Macro-perspective Business knowledge Business commentary Soft skills 14

16 Introducing R

17 Overview of R Profile R- Statistical Programing Language Developed in 1994 Originated at the University of Auckland, New Zealand Created by Ross Ihaka & Robert Gentleman 2 Millions Users shown in Blue from the Map 16

18 Overview of R 2 Millions R users 17

19 Why R? Free Open Source Package Ecosystem Learning curve Strong Community Memory Management Visualization 18

20 R has 9,153 packages built and ready for use If you re trying to do something that s not in the code set, you go out and find an R package... and then you snap it right in and start using it, -Robert Sudol, Sr. Development Manager in Fixed Income Technology, AllianceBernstein. 19

21 Predictive modelling in R 20

22 Graphics in R 21

23 Getting started

24 Working in R-Studio 23

25 Get practicing Use existing datasets in R and run a statistics summary output Call existing datasets Look up info datasets Select dataset called AirPassangers Summary Stats 24

26 Useful packages for Actuaries Actuar Actuarial Functions and Heavy Tailed Distributions ActuDistns Functions for actuarial scientists CompLognormal Functions for actuarial scientists ChainLadder Statistical Methods and Models for Claims Reserving in General Insurance Lifecontingencies Financial and Actuarial Mathematics for Life Contingencies Raw R Actuarial Workshops Tweedie Tweedie exponential family models which is useful for modeling pure premiums. insurancedata A Collection of Insurance Datasets Useful in Risk Classification in Non-life Insurance 25

27 Useful Functions in R Function Syntax Descriptions/Results Help Help.start() User manuals?hist How to produce histograms Maintenance ls() gc() Memory.size() Find out what objects are in memory Clean up the memory and check how much storage available Calculations Vectors X<- -5 abs(x) X X1 <- C(1,2,3) X1 Absolute value of -5 is 5. <- means assign value to X [1]

28 Demo in R Studio

29 Act now and gain a competitive advantage of using R 28

30 R U up on R? Society of Actuaries Annual Meeting Las Vegas 25-OCT :15 11:30 am By Dave Snell, ASA, MAAA, CLU, ChFC, FLMI, ACS, ARA, MCP Technology Evangelist RGA 25-OCT-2016

31 Why Learn Yet Another Language? Actuaries who want to stay viable in the data analysis space need to upgrade their skill sets beyond just spreadsheets. Data is getting BIGGER! R is one (of many) new tools for data analysis and presentation. gigabytes terabytes petabytes exabytes zettabytes yottabytes brontobytes geopbytes oh, my! The SOA is adding predictive analytics to the ASA syllabus 2

32 Big Data is all around us much publicly posted 1% sample of 332,900 tweets in 5 seconds > proc.time()-ptm user system elapsed > > tweets.df <- parsetweets("tweets_sample.json") tweets have been parsed. > tail(tweets.df$text,20) [1] ไม เห นด วยก บร ปน เลย ไม ใช คนไทยท กคนท ค ดแบบน แล วก ไม ใช ฝร งท กคนท ค ดแบบน คนไทยด ๆก ม ฝร งแย ๆก ม " [2] "Psychedelic Padded Pipe Pouch by #hippy #weed #smoke #can [3] WISCONSIN,, TRUMPSTERS, AMERICANS, WE COME TOGETHER FOR A BATTLE TODAY, AND FOR OU [4] "@tabo_luv_st 音だけ流れ続けて画面真っ暗 ~www" [5] "@nozomieiei 知ってる " [6] "So much pain inside him.immense betray from Yulin humans #StopYuLin4ever [7] Check out these headphones!! #essential5sos [8] et le 3x01 qd il l'appel pr son anniv alors qu'il a perdu son humanité [9] " 猫をあやす棗さんが可愛すぎて歯磨き粉噴出した " [10] " こんな時間に腹減り " [11] 大変だった時に使うハンコできた [12] " あっ " [13] " モイ!iPhoneからキャス配信中 - [14] พ คโบกอม ถ ายแบบให ก บแบรนด MontBell คอลเลคช น S/S 2016 / หล อ น าร ก \xed \xed \u [15] ไม ใช คนท โลกส วนต วส งคร บ ไม ใช คนท เข าก บคนยาก ตรงก นข ามผมเข าก บคนอ นง าย แต ผมแค เล อกคนท จะให ร เร องส วนต วของ " [16] 青 パクに土偶と埴輪の違いは解りますか? って聞いてみたら \n 緑 解りますよ! 土偶はこう ( 土偶のポーズ ) で埴輪はこう ( 埴輪のポーズ ) ですよね! って答えられた そういう話じゃない " [17] don't worry. Even in Japan I wouldn't have done that. What do you take me for?? Some weeb?? [18] "Ladies [19] " 定期 すべての人に好かれる気はないし必要ないと思ってる ごく少数の仲のいい人が出来ればそれでいい " [20] "@june7845 고양이귀랑꼬리랑발달고고양이란제리랑스타킹입고사진찍자 " 3

33 How will they dramatically change the future of health insurance? The internet of things will know more about you than any personal doctor could ever hope to know about you. Wearables; watches, shirts, socks, etc. Embeddables: pills, nanobots, labs in your bloodstream Appliances: smart fridge, lav results, Kindle reading, movies and shows watched Consumables: the telltale hamburger, bragging broccoli These go beyond Big Brother s wildest dreams! 4

34 How are Big Data and predictive analytics changing healthcare? The Truman Show was just the Beginning! Genome Phenome Physiome Anatome Transcriptome Proteome Metabolome Microbiome Epigenome Exposome Try but be very afraid! A Panomic perspective! 5

35 So, why R, when there are so many tools for predictive analytics? Free (instead, spend $25 to join the Predictive Analytics and Futurism section) Now more popular than SAS Easier for statisticians than Python Open Source (easier for others to make packages for you) Thousands of package already built and documented Free no licensing issues MatLab costs a lot of money Millions of programmers seems to be gaining momentum Supportive community online to help you get over obstacles Lots of free and readily available tutorials and examples Runs on most platforms (Windows, ios, Linux, etc.) Great graphics capability (especially via gglot2) Free OK to copy and share with your friends 6

36 Heresy: I am not recommending that you start with R-Studio even though it is great. Home screen of Jupyter.org Get instructions for installing R with Jupyter at 7

37 One of the best ideas I got from the Johns Hopkins courses was the importance of codebooks. 8

38 R differs (from other languages) in the assignment syntax Assignment of values to variables: X = 5, X <- 5, 5-> X, assign( X,5) are identical There are four ways to assign a value to a variable: X=5 requires the least typing and is easily read by most folks familiar with other programming languages X<-5 appeals to mathematicians, who always objected to the equals sign for assignment because of statements like x=x+1 5->x is another step towards clarity (put 5 into the variable x) but it is cumbersome when the left side is a long formula Assign( x,5) satisfies the purists; but involves the most typing. It is handy for generating dynamic code programmatically. Bottom line: choose whatever assignment style you wish, but be prepared to read it in any of the four formats. The convention seems to be X <- 5 for a variable and X=5 for a parameter 9

39 Quotes can be or but be consistent Single or double quotes can be used to enclose strings. This allows you to use them in strings. A<- abc, B= abc, C<- doesn t cause error, D= it is OK to include quotes in strings R is case sensitive: ABC, abc, Abc, abc, abc, ABc, AbC, abc are eight different variables. most common variable types: numeric (5.3, 7, pi), character ( a string, a string ), Boolean (TRUE, FALSE, T, F) to see type, use class(x)[1] "numeric" to test type, use is.numeric(x), is.character(x), is.boolean(x), etc. 10

40 A few more tips: Be careful; with the = assignment operator x=10 assigns 10 to x but x == 10 tests to see if x equals 10 Useful functions : getwd() #get working directory[1] "C:/Users/Dave/Documents" ls() #lists all objects currently defined "loc" "num" "rules" "string" "system" "variables" "x" "X" rm(num) #removes the object num from memory ls() "loc" "rules" "string" "system" "variables" "x" "X" rm(list=ls()) #removes all objects from memory ls() character(0) 11

41 Quick demo of R in a JuPyteR notebook The demo will show very short code segments that make (live demo!) a variety of interesting types of charts, interact with an Excel workbook, and make documentation of your programs a lot less boring. You will get a link to the source code so you can experiment at your leisure afterwards. The charts are from the R Core Team They have supplied many free tuturials Get instructions for installing R with Jupyter at 12

42 Why R? It s a primary language for predictive analytics You have a supportive community online to help you get over obstacles There are lots of free and readily available tutorials and examples It runs on most platforms (Windows, ios, Linux, etc.) It offers great graphics capability (especially via gglot2) It s easy to spell You might just have fun with it! 13

43 R U up on R? Society of Actuaries Annual Meeting Las Vegas 25-OCT :15 11:30 am By Dave Snell, ASA, MAAA, CLU, ChFC, FLMI, ACS, ARA, MCP Technology Evangelist RGA 25-OCT-2016