r we fucked? oscon.pdf · who uses r? • r is used by anyone who needs visualize or extract...
TRANSCRIPT
Rwef#¢$ed?
WhyWeDon'tUnderstandRisk,andHowitDoomsUsAll
StrangerswithCandy?
• Yoursix‐year‐olddaughterislostinthecity.Whatshouldshedo?
www.flickr.com/photos/keyphotographics/2185720078
Seventhingsyoushouldknowaboutrisk
Risk
=Probability×Impact
1:
Models:notjustfordressingupandfallingdown
www.flickr.com/photos/edgun/2624354242
2:Allmodelsarewrong,butsomeareuseful.
G.E.P.Box,Sta_s_cian
Housingpricesonlyevergoup!
Er,well…
RobertShiller,www.irra_onalexuberance.com
#3:Coincidencesarenotsurprising
JessicaUcs:hcp://anson.ucdavis.edu/~ucs/AmerStat2003.pdf
Thesethingscouldneverhappen!
• Twosubmarinescrashintheocean• Meteorstrikesanairplane
• Cosmicraysmakeacomputa_onfail
“Everythingweseehasaboutazeroprobability.Calcula_ngtheseprobabili_esaherthefactiskindofmeaningless.”– PeterWesjall
#4:Condi_onalprobability:you’redoingitwrong
JessicaUcs:hcp://anson.ucdavis.edu/~ucs/AmerStat2003.pdf
Talkingonyourcellphonewhiledrivingistotallysafe
AAA(2001):– 1.5%ofdriversinaccidentswereusingacellphone
– 10.9%ofdriversweredistractedbyapassenger– OMG!Passengersaremoredangerousthancellphones?
#5:Surveysandexperimentsprobablyhavebias
“Sureofficer,I’lltellyouexactlywhatadumbassIjustwas.”
• Furthercri_quebyMagliozzi&Magliozzi(2001):thisisvolunteeredinforma_on
• hcp://_nyurl.com/R‐aaa‐study
©Pixar,Inc
#6:Thepluralofanecdoteisnotdata
FrankKotsonis
Globalwarmingistotallybunk!
hcp://flowingdata.com/2009/07/20/important‐data‐please‐act‐responsibily/
#7:Correla_ondoesn’timplycausa_on
Sugges_vely‐wagglingeyebrows
hcp://xkcd.com/552/
Open‐sourcesta_s_calanalysisandvisualiza_on
WhatcanyoudowithR?
• Improveyourcareerprospects
Dice.com: "Is There an R in Your Future?” h,p://0nyurl.com/R‐dice‐video
WhatisR?
• Risopen‐sourcesohwareforsta_s_calcompu_nganddatavisualiza_on.
• Risanextensible,interac_veprogramminglanguage.
• Rhasavastlibraryofmathema_calandsta_s_calalgorithms.
• Rhasaworld‐widecommunityof1‐2millionusersanddevelopers.
• RrunsonWindows,Mac,LinuxandUnixsystems.
WhousesR?
• Risusedbyanyonewhoneedsvisualizeorextractinforma_onfromdata:– sta_s_cians,scien_sts,dataanalysts,dataminers,quan_ta_veanalysts,modelers,forecasters
• Risusedtoanalyzeandpresentdataatorganiza_onslike:– Google,BankofAmerica,Pfizer,Facebook,Shell,theNewYorkTimes,theFoodandDrugAdministra_on
• Risusedasafreealterna_vetoproductslike:– SAS,SPSS,Stata,Sta_s_ca,S‐PLUS.
• Risusedbyacademicstoimplementthelatestadvancesinsta_s_csandpredic_veanaly_cs
Rsaves_mefortheNewYorkTimes
• Published3hoursaherJackson’sdeath:
nyt.com, June 25 2009
WhatcanyoudowithR?
• Mash‐upmessydatasourcestoanalyzetheforeclosurecrisis
From O'Reilly's Data Mash-ups in R.
WhatcanyoudowithR?
• FindacleanplacetosurfintheBayArea
John Oram, a scientist at the San Francisco Estuary Institute (SFEI) uses R to collect and monitor environmental data from the waters and wetlands of the Bay Area
WhatcanyoudowithR?
• Comparebaseballplayerperformance
PitchFX Viewer, by Mike Driscoll: labs.dataspora.com/gameday/
WhatcanyoudowithR?
• BuildaWebdashboard,fast.
From the "Learning R" blog, "R: From zero to Web 2.0 in six weeks"
WhatcanyoudowithR?
• Showthata_ppingpointingaymarriagesupportisnear.
The chart is based on a hierarchical model of polling data by Jeff Lax and Justin Philips using the glmer function in R
Let’sdoasimula_on!
• Isthisyourbirthday?
6
January
Simula_ngbirthdays
• Asimplesimulation:
birthday <- function(n) { ntests <- 10000 pop <- 1:365 anydup <- function(i) any(duplicated( sample(pop, n, replace=TRUE))) sum(sapply(seq(ntests), anydup)) / ntests }
x <- foreach (j=1:100) %dopar% birthday (j)
BirthdaySimula_on
> x <- foreach (j=1:100) %dopar% birthday (j) > plot(1:100, unlist(x),type="l")
ThankYou!
• DavidSmith– david@revolu_on‐compu_ng.com,@revodavid
• DaneseCooper– danese@revolu_on‐compu_ng.com
• REvolu_onCompu_ng– Booth405:FreeMonkeysat6PM! – www.revolu_on‐compu_ng.com
• Revolu1ons,theRblog– blog.revolu_on‐compu_ng.com
• RProject– www.r‐project.org