Using R in Insurance GIRO 2012

Using_R_in_Insurance_GIRO_2012.Rmd

Using R in Insurance
==============================

```{r results='asis', echo=FALSE, message=FALSE}

library(ChainLadder)
library(googleVis)
data(RAA) # example data set of the ChainLadder package
class(RAA) <- "matrix" # change the class from triangle to matrix
df <- as.data.frame(t(RAA)) # coerce triangle into a data.frame
names(df) <- 2002 : 2011
df$dev <-  1:10
LC <- gvisLineChart(df, "dev", options=list(gvis.editor="Edit me!", 
                                            title="Incurred claims",
                                            hAxis='{title:"Development year"}',
                                            width=600, height=350))
print(LC, 'chart')
```
Markus Gesmann, GIRO Brussels 
19 September 2012


# Hello - About me 
<ul> 
  <li>Name: [Markus Gesmann](https://plus.google.com/118201313972528070577/posts)</li>
  <li> Profession: Mathematician, working as an analyst at [Lloyd's](http://www.lloyds.com)</li>
  <li>Maintainer and co-author of two R packages:
  <ul>
  <li>[ChainLadder](http://code.google.com/p/chainladder/): Statistical methods for the calculation of outstanding claims reserves in general insurance</li>
  <li> [googleVis](http://code.google.com/p/google-motion-charts-with-r/): Interface between R and the Google Visualisation API</li>
  </ul>
  </li>
  <li> Blogger: [mages' blog](http://lamges.blogspot.com)</li>
</ul>

# Agenda 

<ul>
<li>My motivation for using R</li>
<li>Brief history of R and its way into insurance</li>
<li>Three R examples</li>
<li>R for actuaries: Where to start?</li>
<li>How I created this presentation with R</li>
<li>Conclusions and discussion</li>
</ul>

# Why I started using R
<ul>
<li> I started in insurance in 2003, fresh out of university</li>
<li> I had used a variety of software, tools and languages already</li>
<li> But I had no experience with spreadsheets</li>
<li> I was surprised how people worked in insurance</li>
<li> I was really surprised what colleagues did with spreadsheets</li>
<li> I looked for alternative data analysis tools</li>
<li> All the cool kids were talking about R</li>
<li> I wanted to be cool as well</li>
</ul>
<a href="http://lamages.blogspot.com">
<img src="http://2.bp.blogspot.com/-NAMnviVPFnw/Tw_OJRa7kII/AAAAAAAAAIo/ERYyvM2dV84/s1600/photo-1.JPG" alt="mages blog" /></a>


# What is R?

<img src="http://www.r-project.org/hpgraphic.png" alt="R project" />
<ul>
  <li> <a href="http://www.r-project.org">R</a> is a free software environment for statistical computing and graphics </li>
  <li> It compiles and runs on a wide variety of UNIX platforms, Windows and OS X</li>
  <li> <a href="http://www.burns-stat.com/pages/Present/infernoishR_annotated.pdf">Notes on the history of R by Pat Burns</a></li>
</ul>

# R made its way from academia to industry
<ul>
  <li> R started at universities in the <a href="http://www.stat.auckland.ac.nz/~ihaka/downloads/R-paper.pdf">1990s</a></li>
  <li> It is still largely maintained by academics</li>
  <li> R is widely used for research publications</li>
  <li> <a href="http://lamages.blogspot.co.uk/2011/10/r-related-books-traditional-vs-online.html">Over 100 text books with R have been published already</a></li>
  <li> Today many graduates leave universities with some R knowledge </li>
</ul>

# R made its way from academia to industry
<ul>
<li> Many commercial applications support R: 
  <ul>
  <li><a href="http://www.oracle.com/us/corporate/features/features-oracle-r-enterprise-498732.html">Oracle</a>, <a href="http://help.sap.com/hana/hana_dev_r_emb_en.pdf">SAP Hana</a>, <a href="http://support.sas.com/rnd/app/studio/Rinterface2.html">SAS</a>, <a href="http://www-01.ibm.com/software/analytics/spss/products/statistics/developer/">IBM SPSS</a>, <a href="http://spotfire.tibco.com/~/media/content-center/datasheets/r-splus.ashx">Spotfire</a>, <a href="http://www.statconn.com">MS Office</a>, <a href="http://community.qlikview.com/docs/DOC-2975">QlikView</a>, etc.
  </li>
  </ul></li>
  <li> Commercial support is available from third parties, e.g.
  <ul>
  <li><a href="http://www.mango-solutions.com">Mango Solutions</a>, <a href="http://www.revolutionanalytics.com">Revolution Analytics</a>, <a href="http://www.trinostics.com">Trinostics LLC</a>, <a href="http://www.burns-stat.com">Burns Stats</a></li>
 <li>Or contact me</li>
 </ul></li>
  <li> R is established in many disciplines outside academia, e.g. <a href="http://cran.r-project.org/web/views/ClinicalTrials.html">pharma</a> and <a href="http://cran.r-project.org/web/views/Finance.html">finance</a></li>
  <li> The insurance industry is adopting R as well, e.g. <a href="https://docs.google.com/open?id=0By35Mtg9R9_RZTk5NzM1NGItYmUzMi00MmQ2LTk0MWYtYTY4YzZiNjg0ODc2">Lloyd's</a></li>
</ul>


# Why R in insurance?

<ul>
<li> Why Excel, SAS, SQL, SPSS, Minitab, ...? </li>
<li> [John D. Cook](http://www.johndcook.com/blog/): Why and how people use R
<div align="center">
<video controls="" height="240" poster="http://ch9files.blob.core.windows.net/ch9/3f13/009c002a-e4ef-4aeb-95de-e7ecda173f13/LangNEXT2012JohnCookRLanguage_Custom.jpg" width="320"><source src="http://ak.channel9.msdn.com/ch9/3f13/009c002a-e4ef-4aeb-95de-e7ecda173f13/LangNEXT2012JohnCookRLanguage_mid.mp4" type="video/mp4"></source><source src="http://ak.channel9.msdn.com/ch9/3f13/009c002a-e4ef-4aeb-95de-e7ecda173f13/LangNEXT2012JohnCookRLanguage.webm" type="video/webm"></source></video></div></li>
<li>Because it gets the job done!</li>
</ul>

# But why use a computing language?

<img src="https://lh5.googleusercontent.com/-0mb8ktQfQjE/TwU72nysQaI/AAAAAAAAI1g/EVOnQXhQsuM/s762/geeks-vs-nongeeks-repetitive-tasks.png" alt="Geeks vs Non-Geeks" />

By <a href="https://plus.google.com/102451193315916178828/posts/MGxauXypb1Y">Bruno Oliveira</a>


# Typical use cases for R in insurance 

<ul>
<li> Data transformation</li>
<li> Data analysis</li>
<li> Statistical modelling </li>
<li> Prototyping / ad-hoc work </li>
<li> End user computing </li>
<li> Background statistical engine for applications, e.g. pricing spreadsheet</li>
<li> Reporting and reproducible analysis, e.g. MI, Solvency II documentation  </li>
<li> Learning statistical and actuarial skills </li>
</ul>

# Three R examples
<ul>
<li>Reserving: Mack chain-ladder</li>
<li>Automated reporting: Create PowerPoint slide with R output</li>
<li>Extracting data from a web page: Display earth quakes of the last 30 days</li>
</ul>

# Reserving: Mack chain-ladder

```{r}
library(ChainLadder)
RAA ## Example triangle
```

# Reserving: Mack chain-ladder

```{r results='asis', echo=FALSE, message=FALSE}
RAA2 <- RAA # example data set of the ChainLadder package
class(RAA2) <- "matrix" # change the class from triangle to matrix
df <- as.data.frame(t(RAA2)) # coerce triangle into a data.frame
names(df) <- 2002 : 2011
df$dev <-  1:10
LC <- gvisLineChart(df, "dev", options=list(gvis.editor="Edit me!", 
                                            title="Incurred claims",
                                            hAxis='{title:"Development year"}',
                                            width=800, height=500))

print(LC, 'chart')
```

Chart created with <a href="http://code.google.com/p/google-motion-charts-with-r/">googleVis</a>

# Reserving: Mack chain-ladder

```{r}
M <- MackChainLadder(RAA, est.sigma="Mack")
M
```


# Reserving: Mack chain-ladder
```{r eval=FALSE}
plot(M, lattice=TRUE)
```
<img src="http://1.bp.blogspot.com/-0dYQffHdShk/Tr7mZibyQGI/AAAAAAAAAEo/hjb7XhrO_Mc/s1600/MackLattice.png" alt="Mack chain ladder output" />

# Automated reporting: Create PowerPoint slide with R output
```{r eval=FALSE, tidy=FALSE}
myfile=tempfile()
win.metafile(file=myfile)
plot(M, lattice=TRUE)
dev.off()
## Load MS Office interface
library(rcom)
## Run VBA code from R
ppt<-comCreateObject("Powerpoint.Application")
comSetProperty(ppt,"Visible",TRUE)
myPresColl<-comGetProperty(ppt,"Presentations")
myPres<-comInvoke(myPresColl,"Add")
mySlides<-comGetProperty(myPres,"Slides")
mySlide<-comInvoke(mySlides,"Add",1,12)
myShapes<-comGetProperty(mySlide,"Shapes")
myPicture<-comInvoke(myShapes,"AddPicture",
                     myfile, 0,1,100,10)
```


# Extracting data from a web page 

```{r results='asis', tidy=FALSE}
library(XML)
library(googleVis)
## Source data diretly from the web
url <- "http://www.iris.edu/seismon/last30.html"
eq <- readHTMLTable(readLines(url),
 colClasses=c("factor", rep("numeric", 4), "factor"), which=2)
## Format location data
eq$loc=paste(eq$LAT, eq$LON, sep=":")
```
<div align="center">
```{r results='asis' , echo=FALSE, message=FALSE}
tbl <- gvisTable(eq, options=list(width=800, height=250))
print(tbl, 'chart')
```
</div>

# Display earth quake information of last 30 days
```{r tidy=FALSE}
library(googleVis)
## Create a geo chart with the Google Chart API
G <- gvisGeoChart(eq, "loc", "DEPTH km", "MAG",
                   options=list(displayMode="Markers", 
                   colorAxis="{colors:['purple', 'red', 'orange', 'grey']}",
                   backgroundColor="lightblue"), chartid="EQ")

```
```{r eval=FALSE}
plot(G)
```
<div align="center">
```{r results='asis' , echo=FALSE, message=FALSE}
print(G, 'chart')
```
</div>

# Getting started with R in actuarial work

<ul>
<li><a href="http://toolkit.pbwiki.com/f/R%20Examples%20for%20Actuaries%20v0.1-1.pdf">Introduction to R for Actuaries</a> by Nigel de Silva</li>
<li><a href="http://www1.fee.uva.nl/ke/act/people/kaas/ModernART.htm">Modern actuarial risk theory using R</a> by Kaas, Goovaerts, Dhaene and Denuit</li>
<li><a href="http://www.slideshare.net/dataspora/an-interactive-introduction-to-r-programming-language-for-statistics">An Interactive Introduction To R</a> by Michael Driscoll and Dan Murphy</li>
<li><a href="http://www.favir.net/">Formatted Actuarial Vignettes in R</a> by Ben Escoto</li>
<li><a href="http://www.actuaries.org.uk/system/files/documents/pdf/actuarial-toolkit.pdf">An Actuarial Toolkit</a> presented at GIRO convention 2006 in Vienna</li>
<li><a href="http://lamages.blogspot.co.uk/2011/09/r-and-insurance.html">Using R at Lloyd's</a> poster at UseR! conference 2011 in Warwick</li>
<li><a href="http://wwww.r-bloggers.com">R-Bloggers</a></li>
</ul>

# R packages for actuaries on CRAN

<ul>
<li><a href="http://cran.r-project.org/web/packages/actuar/index.html">actuar:</a> Loss distributions modelling, risk theory (including ruin theory), simulation of compound hierarchical models and credibility theory</li>
<li><a href="http://cran.r-project.org/web/packages/ChainLadder/index.html">ChainLadder:</a> Reserving methods in R</li>
<li><a href="http://cran.r-project.org/web/packages/copula/index.html">copula:</a> Multivariate Dependence with Copulas</li>
<li><a href="http://cran.r-project.org/web/packages/cplm/index.html">cplm:</a> Monte Carlo EM algorithms and Bayesian methods for fitting Tweedie compound Poisson linear models</li>
<li><a href="http://cran.r-project.org/web/packages/evir/index.html">evir:</a> Extreme Values in R</li>
<li><a href="http://cran.r-project.org/web/packages/fitdistrplus/index.html">fitdistrplus:</a> Help to fit of a parametric distribution to non-censored or censored data</li>
<li><a href="http://cran.r-project.org/web/packages/lifecontingencies/index.html">lifecontingencies:</a> Package to perform actuarial evaluation of life contingencies</li>
<li><a href="http://cran.r-project.org/web/packages/lossDev/index.html">lossDev:</a> A Bayesian time series loss development model</li>
<li><a href="http://cran.r-project.org/web/packages/mondate/index.html">mondate:</a> R package to keep track of dates in terms of months</li>
</ul>


# Meet the R experts

<table>
<tr>
<td>
<ul>
<li><a href="https://stat.ethz.ch/mailman/listinfo/r-sig-insurance">R special interest group insurance email list</a> </li>
<li> <a href="http://www.londonr.org">London R user group</a></li>
<li> <a href="https://www.rmetrics.org">R/Rmetrics Meielisalp Workshop on Computational Finance and Financial Engineering</a></li>
<li> <a href="http://www.rinfinance.com">R in Finance, Chicago</a></li>
<li> <a href="http://www3.uclm.es/congresos/useR-2013/">UseR! 2013, University of Castilla-La Mancha, Spain</a> </li>
</ul>
</td><td>
<img height="350px" src="http://photos1.meetupstatic.com/photos/event/7/a/highres_133320122.jpeg" alt="London R photo" />

London R user group meeting
</td>
</tr>
</table>



# How I created this presentation with RStudio, knitr, pandoc and slidy  
<ul>
<li> [knitr](http://yihui.name/knitr/) is a package by [Yihui Xie](http://yihui.name/) that brings literate programming to a new level 
  <ul>
  <li>It allows to create content really quickly, without worrying to much about layout and R formatting</li>
</ul></li>
<li> [RStudio](http://rstudio.org) integrated knitr into its IDE, which allows to knit Rmd-files by the push of a button into markdown</li>
<li> Markdown output can be converted into several other file formats, such as html, with [pandoc](http://johnmacfarlane.net/pandoc/)</li>
<li> [slidy](http://www.w3.org/Talks/Tools/Slidy2/Overview.html) is one of the options to create interactive html-slides with pandoc</li>
<li> For more details see my recent [blog post](http://lamages.blogspot.co.uk/2012/05/interactive-reports-in-r-with-knitr-and.html) and [source code of this talk](https://gist.github.com/3687713).</li>
</ul>
<code><pre>
Rscript -e "library(knitr); knit('Using_R_in_Insurance_GIRO_2012.Rmd')"
pandoc -s -S -i -t slidy --mathjax Using_R_in_Insurance_GIRO_2012.md 
  -o Using_R_in_Insurance_GIRO_2012.html
</pre></code>


# Conclusions

<ul>
<li> R comes with lots of functions for actuarial work</li>
<li> It provides an ideal framework for end user computing</li>
<li> The momentum behind R has grown significantly over the last 5 years</li>
<li> Today R is often known by graduates - open up to their ideas</li>
<li> Many other software products developed R interfaces </li>
<li> New business models have evolved and will evolve</li>
</ul>

# If you liked this presentation ...
... you may also like:
<ul>
<li><a href="http://lamages.blogspot.co.uk/2012/09/interactive-web-graphs-with-r-overview.html">Interactive web graphs with R - Overview and googleVis tutorial</a>, Royal Statistical Society Conference, 2012</li>
<li><a href="http://www.rinfinance.com/agenda/2012/talk/MarkusGesmann.pdf">Overview of Lloyd's using R and googleVis</a>, R in Finance, 2012</li>
<li><a href="http://chainladder.googlecode.com/files/ChainLadder_Markus_20010Nov10.pdf">ChainLadder at the Predictive Modelling Seminar</a>, Institute of Actuaries, 2010 </li>
<li><a href="http://code.google.com/p/chainladder/downloads/detail?name=R_and_MS_Office-MG-20100504.pdf">How to integrate R into MS Office</a>, LondonR, 2010 </li>
<li><a href="http://lamages.blogspot.co.uk/2011/12/fitting-distributions-with-r.html">Fitting distribution with R</a></li>
<li><a href="http://lamages.blogspot.co.uk/2012/01/say-it-in-r-with-by-apply-and-friends.html">Say it in R with "by", "apply" and friends</a> </li>
</ul>

# Questions?

<ul>
<li> Idea: R in Insurance Workshop - Interest? </li>
<li>Contact: markus dot gesmann at gmail dot com</li>
</ul>