122S:166Computing in StatisticsIntroductionLecture 1August 22, 2011Kate Cowles374 SH, [email protected] endeavors• three branches– applied statistics and data analysis– development of stati stical methods and soft-ware– research in statistical theory• computing essential to all of themI keep saying that the sexy job in thenext 10 years will be statisticians. AndI’m not kidding.Hal Varian, chief economist at Google (New Yo rkTimes, Aug. 6, 2009)http://www.nytimes.com/2009/08/06/technolog06stats.html3Goals of this course are to develop:• intelligent use of appropriate computing toolsfor both statistical endeavors– R/Splus– SAS– database management software and con-cepts• understanding of important statistical com-puting algorithms– Newton’s method– EM al g orithm– the bootstrap• ability to design and implement simulationstudies4• communication of statistical ideas in words,numbers, and graphics– LATEX– format of scientific reporting5Types of computer productsThis section and the next 2 borrow heavily from Chapter 1of the course notes for “Statistical Computing and Graph-ics” by Frank Harrellhesweb1.med.virginia.edu/biostat/teaching/statcomp• operating systems: make the computer itselfworke.g. Linux, Windows, Unix, MacOS• applications: perform specific taskse.g. Microsoft Word, Excel, S-Plus, OpenO f-fice, R, SAS, . . .• commercial systems– code and lists of bugs are secret– expensive– require upgrading and relicensing– Microsoft products, S-Plus, SAS, SPSS,Unix, etc.• free O pen Source systems6– revolution in software availabil ity and f unc-tion from the open source movement– can see all code, change it, learn from it– quality generally quite good∗ often better than that of commercially-developed software because Open Sourcesoftware has been tested by more peopleunder more different co nditions– more rap id updates– most products have an active and helpfuluser news group– generally lack some fancy features like ex-tensive GUI– Linux, LATEX , R7User interfaces: graphi cal vscommand line• graphical (GUI, mouse, menus)– easier to learn– less flexi ble– repetitive when the same tasks have to berepeated– hard to document the exact steps taken– hard to reproduce results• command line interfaces– harder to learn– more flexib le and powerful– can save commands in scripts to replaywhen the same tasks h ave to be performedrepeatedly– can write generic commands to facilitaterunning different analyses with the samestructure8Types of user files• text• binary• graphics files9Linux historyThe material in this section borrows heavily from Section 1.1of Introduction to Linux: A Hands on Guide by MachteltGarrels.http://www.tldp.org/LDP/intro-linux/intro-linux.pdf• Unix– 1969: team of developers at Bell Labora-tories began work on solution to problemof software incompatibility∗ at that time, every model of co mputerhad different operating system∗ software was customized to specific pur-poses, and ran on only one type of com-puter system– UNIX operating system n ee ded only smallpiece of cod e specific to one type of com-puter: the kernel– operating system (and all other functi ons)built around kernel10– higher-level programming language C spe-cially developed for creating UNIX– at first used only in very large comput-ing environments — universities, govern-ment, large corporatio ns with mainframesand minicomputers• developments in 1970’ s and 1980’s– continued development of UNIX– support of UNIX in products of increa si ngnumbers of ha rdware an d so ftware vendors– invention of personal computers– by end of 1980’s, several versions of UNIXavailable for PC architecture, but not free11• Linus Torvalds and Linux– computer science student at University ofHelsinki– goal: to create a freely-available operatingsystem that was compliant with originalUNIX– began working on it in early 1990’s– other coders jumped aboard to developdrivers to make Linux usable with moreand more hardware– 12000 Linux users by 1993– all features of UNIX added over few moreyears12• Linux to day– only operating system in the world thatruns on as wide a range of hardware∗ desktop workstations∗ mid- a nd high-end servers∗ PDAs, netb oo ks, experimental wristwatches,etc.– well known as a stable and reliab le plat-form for servers– examples of users∗ Amazon (Internet book seller)∗ United States Post Office∗ German a rmy∗ high-energy p hysics grid13Logging in, activating the userinterface, and logging out• PC-based Linux systems have two basic modes:graphical an d text• in gra phical mode (usually the de fa ult)– login requires i nputting user name in onewindow and password in anothe r– make sure mouse po inter is in the l o g inwindow; press Enter after entering the user-name and after entering the password– open a terminal window by left clickingon icon of a compu ter screen a t bottom ofscreen– log out by∗ closing all terminal windows and appli-cations∗ clicking “System” at bottom of screenand selectin g “Log Out”14– in text mode∗ whole screen is black, with white char-acters∗ you are prompted for username and pass-word; press enter after typin g in eachone∗ log out by entering logout commandand pressing enter15Linux student computer labs• UI Mathematical Sciences educational com-puter la bo ra to ri es:– 346 S H (primarily for statistics students)– B5 MLH and 301 MLH (primarily for mathand c .s.)• your username an d pa ssword are assigned byComputer Support Group– the first time you log in, you should im-mediately change the password they giveyou– enter the passwd command in a terminalwindow and follow the instructio ns– requirements for passwords∗ from six to eight characters long∗ use b oth uppercase and lowercase let-ters and either digits or non-alpha beti csymbols (or both)16∗ pick somethin g you will remember with-out writi ng it down∗ don’t use the same password on differentsystems• link to information from the Computer Sup-port Group (CSG) on student a ccounts onthe math/c. s. / stati stics networkwww.divms.uiowa.edu/help/orientation• must bring your own printer paper if youwish to print• to access Internet, click Firefox icon (worldwith a flaming fox) th e bo ttom of the screen• to log out, click “System” and select
View Full Document
Unlocking...