{"id":762,"date":"2015-02-17T18:14:23","date_gmt":"2015-02-18T01:14:23","guid":{"rendered":"http:\/\/homepages.uc.edu\/~yaozo\/wordpress\/?p=762"},"modified":"2015-02-17T18:14:23","modified_gmt":"2015-02-18T01:14:23","slug":"10-r-packages-which-will-master-you-in-r-tool","status":"publish","type":"post","link":"https:\/\/zhuoyao.net\/index.php\/2015\/02\/17\/10-r-packages-which-will-master-you-in-r-tool\/","title":{"rendered":"10 R packages which will master you in R Tool"},"content":{"rendered":"

R can be more prickly and obscure than other languages like Python or Java. The good news is that there are tons of packages which provide simple and familiar interfaces on top of Base R. This post is about ten packages I love and use everyday and ones I wish I knew about earlier.<\/p>\n

\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0<\/strong><\/p>\n

    \n<\/ol>\n
      \n
    1. \n
      install.packages(\"sqldf\")<\/pre>\n
      One of the steepest parts of the R learning curve is the syntax. It took me a while to get over using\u00a0<-<\/code>\u00a0instead of\u00a0=<\/code>. I hear people say a lot of times\u00a0How do I just do a<\/i>\u00a0VLOOKUP<\/code>?!?<\/i>\u00a0R is great for general data munging tasks, but it takes a while to master. I think it\u2019s safe to say that\u00a0sqldf<\/code>\u00a0was my R \u201ctraining wheels\u201d.<\/p>\n
      sqldf<\/code>\u00a0let\u2019s you perform SQL queries on your R data frames. People coming over from SAS will find it very familiar and anyone with basic SQL skills will have no trouble using it\u2013sqldf<\/code>\u00a0uses\u00a0SQLite syntax.<\/a><\/p>\n
      \n
      \n\n\n\n
      <\/td>\n\n
      \n
      \n
      \n
      library(sqldf)<\/em><\/div>\n
      <\/div>\n
      sqldf(\u201cSELECT<\/em><\/div>\n
      day<\/em><\/div>\n
      , avg(temp) as avg_temp<\/em><\/div>\n
      FROM beaver2<\/em><\/div>\n
      GROUP BY<\/em><\/div>\n
      day;\u201d)<\/em><\/div>\n
      <\/div>\n
      # day avg_temp<\/em><\/div>\n
      #1 307 37.57931<\/em><\/div>\n
      #2 308 37.71308<\/em><\/div>\n
      <\/div>\n
      #beavers1 and beavers2 come with base R<\/em><\/div>\n
      beavers <- sqldf(\u201cselect * from beaver1<\/em><\/div>\n
      union all<\/em><\/div>\n
      select * from beaver2;\u201d)<\/em><\/div>\n
      #head(beavers)<\/em><\/div>\n
      # day time temp activ<\/em><\/div>\n
      #1 346 840 36.33 0<\/em><\/div>\n
      #2 346 850 36.34 0<\/em><\/div>\n
      #3 346 900 36.35 0<\/em><\/div>\n
      #4 346 910 36.42 0<\/em><\/div>\n
      #5 346 920 36.55 0<\/em><\/div>\n
      #6 346 930 36.69 0<\/em><\/div>\n
      <\/div>\n
      movies <- data.frame(<\/em><\/div>\n
      title=c(\u201cThe Great Outdoors\u201d, \u201cCaddyshack\u201d, \u201cFletch\u201d, \u201cDays of Thunder\u201d, \u201cCrazy Heart\u201d),<\/em><\/div>\n
      year=c(1988, 1980, 1985, 1990, 2009)<\/em><\/div>\n
      )<\/em><\/div>\n
      boxoffice <- data.frame(<\/em><\/div>\n
      title=c(\u201cThe Great Outdoors\u201d, \u201cCaddyshack\u201d, \u201cFletch\u201d, \u201cDays of Thunder\u201d,\u201dTop Gun\u201d),<\/em><\/div>\n
      revenue=c(43455230, 39846344, 59600000, 157920733, 353816701)<\/em><\/div>\n
      )<\/em><\/div>\n
      <\/div>\n
      sqldf(\u201cSELECT<\/em><\/div>\n
      m.*<\/em><\/div>\n
      , b.revenue<\/em><\/div>\n
      FROM<\/em><\/div>\n
      movies m<\/em><\/div>\n
      INNER JOIN<\/em><\/div>\n
      boxoffice b<\/em><\/div>\n
      ON m.title = b.title;\u201d)<\/em><\/div>\n
      <\/div>\n
      # title year revenue<\/em><\/div>\n
      #1 The Great Outdoors 1988 43455230<\/em><\/div>\n
      #2 Caddyshack 1980 39846344<\/em><\/div>\n
      #3 Fletch 1985 59600000<\/em><\/div>\n
      #4 Days of Thunder 1990 157920733<\/em><\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n
      <\/div>\n
      <\/div>\n<\/div>\n<\/li>\n
    2. \n
      install.packages(\"forecast\")<\/pre>\n
      I don\u2019t do time series analysis very often, but when I do\u00a0forecast<\/code>\u00a0is my library of choice.\u00a0forecast<\/code>\u00a0makes it incredibly easy to fit time series models like ARIMA, ARMA, AR, Exponential Smoothing, etc.<\/p>\n
      \n
      \n\n\n\n
      <\/td>\n\n
      \n
      \n
      \n
      library(forecast)<\/div>\n
      <\/div>\n
      # mdeaths: Monthly Deaths from Lung Diseases in the UK<\/div>\n
      fit <- auto.arima(mdeaths)<\/div>\n
      #customize your confidence intervals<\/div>\n
      forecast(fit, level=c(80, 95, 99), h=3)<\/div>\n
      # Point Forecast Lo 80 Hi 80 Lo 95 Hi 95 Lo 99 Hi 99<\/div>\n
      #Jan 1980 1822.863 1564.192 2081.534 1427.259 2218.467 1302.952 2342.774<\/div>\n
      #Feb 1980 1923.190 1635.530 2210.851 1483.251 2363.130 1345.012 2501.368<\/div>\n
      #Mar 1980 1789.153 1495.048 2083.258 1339.359 2238.947 1198.023 2380.283<\/div>\n
      <\/div>\n
      plot(forecast(fit), shadecols=\u201doldstyle\u201d)<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n
      \u00a0<\/a><\/div>\n<\/div>\n
      My favorite feature is the resulting forecast plot.<\/p>\n
      \"forecast<\/li>\n<\/ol>\n
      <\/h5>\n
      3.install.packages(\u201cplyr\u201d)<\/h5>\n
      When I first started using R, I was using basic control operations for manipulating data (for, if, while, etc.). I quickly learned that this was an amateur move, and that there was a better way to do it.<\/p>\n
      In R, the\u00a0apply<\/code>\u00a0family of functions<\/a>\u00a0is the preferred way to call a function on each element of a list or vector. While Base R has this out of the box, its usage can be tricky to master. I\u2019ve found the\u00a0plyr<\/code>\u00a0package<\/a>\u00a0to be an easy to use substitute for\u00a0split<\/code>,\u00a0apply<\/code>,combine<\/code>\u00a0functionality in Base R.<\/p>\n
      plyr<\/code>\u00a0gives you several functions (ddply<\/code>,\u00a0daply<\/code>,\u00a0dlply<\/code>,\u00a0adply<\/code>,\u00a0ldply<\/code>) following a common blueprint: Split a data structure into groups, apply a function on each group, return the results in a data structure.<\/p>\n
      ddply<\/code>\u00a0splits a data frame and returns a data frame (hence the dd).\u00a0daply<\/code>\u00a0splits a data frame and results an array (hence the da). Hopefully you\u2019re getting the idea here.<\/p>\n
      \n
      \n\n\n\n
      \n
      \n
      \n
      library(plyr)<\/div>\n
      <\/div>\n
      # split a data frame by Species, summarize it, then convert the results<\/div>\n
      # into a data frame<\/div>\n
      ddply(iris, .(Species), summarise,<\/div>\n
      mean_petal_length=mean(Petal.Length)<\/div>\n
      )<\/div>\n
      # Species mean_petal_length<\/div>\n
      #1 setosa 1.462<\/div>\n
      #2 versicolor 4.260<\/div>\n
      #3 virginica 5.552<\/div>\n
      <\/div>\n
      # split a data frame by Species, summarize it, then convert the results<\/div>\n
      # into an array<\/div>\n
      unlist(daply(iris[,4:5], .(Species), colwise(mean)))<\/div>\n
      # setosa.Petal.Width versicolor.Petal.Width virginica.Petal.Width<\/div>\n
      # 0.246 1.326 2.026<\/div>\n<\/div>\n<\/div>\n<\/td>\n
      		<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<\/div>\n
      4.install.packages(\"stringr\")<\/pre>\n
      I find base R\u2019s string functionality to be extremely difficult and cumbersome to use. Another package written by\u00a0Hadley Wickham<\/a>,stringr<\/code>, provides some much needed string operators in R. Many of the functions use data strcutures that aren\u2019t commonly used when doing basic analysis.<\/p>\n
      \"stringr
      \nstringr<\/code>\u00a0is remarkably easy to use. Nearly all of the functions (and all of the important ones) are prefixed with \u201cstr\u201d so they\u2019re very easy to remember.<\/p>\n
      \n
      \n\n\n\n
      <\/td>\n\n
      \n
      \n
      \n
      library(stringr)<\/div>\n
      <\/div>\n
      names(iris)<\/div>\n
      #[1] \u201cSepal.Length\u201d \u201cSepal.Width\u201d \u201cPetal.Length\u201d \u201cPetal.Width\u201d \u201cSpecies\u201d<\/div>\n
      names(iris) <- str_replace_all(names(iris), \u201c[.]\u201c, \u201c_\u201d)<\/div>\n
      names(iris)<\/div>\n
      #[1] \u201cSepal_Length\u201d \u201cSepal_Width\u201d \u201cPetal_Length\u201d \u201cPetal_Width\u201d \u201cSpecies\u201d<\/div>\n
      <\/div>\n
      s <- c(\u201cGo to Heaven for the climate, Hell for the company.\u201d)<\/div>\n
      str_extract_all(s, \u201c[H][a-z]+ \u201c)<\/div>\n
      #[[1]]<\/div>\n
      #[1] \u201cHeaven \u201d \u201cHell \u201c<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n
      \u00a0<\/a><\/div>\n<\/div>\n
      5.The database driver package of your choice<\/a><\/h5>\n
      install.packages(\"RPostgreSQL\")\ninstall.packages(\"RMySQL\")\ninstall.packages(\"RMongo\")\ninstall.packages(\"RODBC\")\ninstall.packages(\"RSQLite\")<\/pre>\n
      Everyone does it when they first start (myself included). You\u2019ve just written an awesome query in your preferred SQL editor. Everything is perfect \u2013 the column names are all\u00a0snake case<\/a>, the dates have the right datatype, you finally debugged the\u00a0\"must appear in the GROUP BY clause or be used in an aggregate function\"<\/code>\u00a0issue. You\u2019re ready to do some analysis in R, so you run the query in your SQL editor, copy the results to a csv (or\u2026God forbid\u2026 .xlsx) and read into R.\u00a0You don\u2019t have to do this!<\/strong><\/p>\n
      R has great drivers for nearly every conceivable database. On the off chance you\u2019re using a database which doesn\u2019t have a standalone driver (SQL Server), you can always use\u00a0RODBC<\/code><\/a>.<\/p>\n
      \n
      \n\n\n\n
      \n
      \n
      \n
      library(RPostgreSQL)<\/div>\n
      <\/div>\n
      drv <- dbDriver(\u201cPostgreSQL\u201d)<\/div>\n
      db <- dbConnect(drv, dbname=\u201dncaa\u201d,<\/div>\n
      user=\u201dYOUR USER NAME\u201d, password=\u201dYOUR PASSWORD\u201d)<\/div>\n
      <\/div>\n
      q <- \u201cSELECT<\/div>\n
      *<\/div>\n
      FROM<\/div>\n
      game_scores;\u201d<\/div>\n
      <\/div>\n
      data <- dbGetQuery(db, q)<\/div>\n
      head(data)<\/div>\n
      #id school game_date spread school_score opponent opp_score was_home<\/div>\n
      #1 45111 Boston College 1985-11-16 6.0 21 Syracuse 41 False<\/div>\n
      #2 45112 Boston College 1985-11-02 13.5 12 Penn State 16 False<\/div>\n
      #3 45113 Boston College 1985-10-26 -11.0 17 Cincinnati 24 False<\/div>\n
      #4 45114 Boston College 1985-10-12 -2.0 14 Army 45 False<\/div>\n
      #5 45115 Boston College 1985-09-28 5.0 10 Miami 45 True<\/div>\n
      #6 45116 Boston College 1985-09-21 6.5 29 Pittsburgh 22 False<\/div>\n
      nrow(data)<\/div>\n
      #[1] 30932<\/div>\n
      ncol(data)<\/div>\n
      #[1] 8<\/div>\n<\/div>\n<\/div>\n<\/td>\n
      		<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n
      \u00a0<\/a><\/div>\n<\/div>\n
      Next time you\u2019ve got that perfect query written, just paste it into R and execute it using\u00a0RPostgreSQL<\/code><\/a>,\u00a0RMySQL<\/code><\/a>,\u00a0RMongo<\/code><\/a>,\u00a0RMongo<\/code><\/a>, or\u00a0RODBC<\/code>. In addition to preventing you from having tens of hundreds of CSV files sitting arround, running the query in R saves you time both in I\/O but also in converting datatypes. Dates, times, and datetimes will be automatically set to their R equivalent. It also makes your R script reproducible, so you or someone else on your team can easily produce the same results.<\/p>\n
      \u00a0<\/a><\/h5>\n
      6.install.packages(\"lubridate\")<\/pre>\n
      I\u2019ve never had great luck with dates in R. I\u2019ve never fully grasped the idiosyncracies of working with\u00a0POSIXs<\/a>\u00a0vs. R Dates. Enterlubridate<\/code><\/a>.<\/p>\n
      lubridate<\/code>\u00a0is one of those magical libraries that just seems to do\u00a0exactly<\/em>\u00a0what you expect it to. The functions all have obvious names like\u00a0year<\/code>,\u00a0month<\/code>,\u00a0ymd<\/code>, and\u00a0ymd_hms<\/code>. It\u2019s similar to\u00a0Moment.js<\/code><\/a>\u00a0for those familiar with javascript.<\/p>\n
      \n
      \n\n\n\n
      <\/td>\n\n
      \n
      \n
      \n
      library(lubridate)<\/div>\n
      <\/div>\n
      year(\u201c2012-12-12\u2033)<\/div>\n
      #[1] 2012<\/div>\n
      day(\u201c2012-12-12\u2033)<\/div>\n
      #[1] 12<\/div>\n
      ymd(\u201c2012-12-12\u2033)<\/div>\n
      #1 parsed with %Y-%m-%d<\/div>\n
      #[1] \u201c2012-12-12 UTC\u201d<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n
      \u00a0<\/a><\/div>\n<\/div>\n
      Here\u2019s a really handy reference card that I found in a\u00a0paper<\/a>. It covers just about everything you might conveivably want to do to a date. I\u2019ve also found this\u00a0Date Cheat Sheet<\/a>\u00a0to be a handy reference.<\/p>\n
      \u00a0<\/a><\/h5>\n
      7.install.packages(\"ggplot2\")<\/pre>\n
      Another Hadley Wickham pacakge and probably his most widely known one.\u00a0ggplot2<\/code>\u00a0ranks high on everyone\u2019s list of favorite R pacakges. It\u2019s easy to use and it produces some great looking plots. It\u2019s a great way to present your work, and there are many resources available to help you get started.<\/p>\n