{"id":1460,"date":"2018-06-21T23:50:31","date_gmt":"2018-06-21T21:50:31","guid":{"rendered":"https:\/\/blogs.fu-berlin.de\/reseda\/?page_id=1460"},"modified":"2018-09-25T10:33:29","modified_gmt":"2018-09-25T08:33:29","slug":"r-crash-course-5","status":"publish","type":"page","link":"https:\/\/blogs.fu-berlin.de\/reseda\/r-crash-course-5\/","title":{"rendered":"R Crash Course Part V"},"content":{"rendered":"

<\/a><\/p>\n

9. Missing values<\/h1>\n

Sometimes observations (rows) in a data frame are incomplete. The correct handling of missing values is especially needed in statistical evaluations. Missing values are indicated by the symbol NA<\/span> (Not Available). Non-possible values (e.g., division by 0) are described by the symbol NaN<\/span> (Not a Number). For test purposes, we artificially define two elements as NA<\/span> in our df<\/span>:<\/p>\n

\r\ndf <- data.frame(\r\n  "name" = c("Ben", "Hanna", "Paul", "Arthur"), \r\n  "size" = c(185, 166, 175, 190)\r\n  )\r\n\r\ndf\r\n##     name size\r\n## 1    Ben  185\r\n## 2  Hanna  166\r\n## 3   Paul  175\r\n## 4 Arthur  190\r\n\r\ndf[1:2, 2] <- NA              # generate NAs\r\ndf\r\n##     name size\r\n## 1    Ben   NA\r\n## 2  Hanna   NA\r\n## 3   Paul  175\r\n## 4 Arthur  190\r\n\r\nmean(df$size)                 # mean with NAs\r\n## [1] NA\r\n\r\nmean(df$size, na.rm=TRUE)     # mean with ignoring NAs\r\n## [1] 182.5\r\n<\/pre>\n
Often it makes more sense to record the observations with missing values and to exclude them from your dataset. For detection we can use the is.na()<\/span> function to create a logical vector. This outputs a TRUE for every NA value, a FALSE for every element present. Using the logical vector we can then index our data frame:<\/p>\n
\r\nis.na(df$size)               # logical vector (TRUE for all rows with NAs)\r\n## [1]  TRUE  TRUE FALSE FALSE\r\n\r\ndf[is.na(df$size), ]         # index using logical vector\r\n##    name size\r\n## 1   Ben   NA\r\n## 2 Hanna   NA\r\n\r\ndf[complete.cases(df), ]     # get all rows without NAs\r\n##     name size\r\n## 3   Paul  175\r\n## 4 Arthur  190\r\n<\/pre>\n
<\/a><\/br><\/p>\n
10. Control Structures<\/h1>\n
Control structures allow better control over the execution of our scripts, e.g., if<\/span>, if else<\/span>, else<\/span>, while<\/span>, switch<\/span>, repeat<\/span>, break<\/span>, return<\/span> …
\nUse the help function for a more detailed documentation. <\/p>\n
The IF statement<\/strong><\/p>\n
Use the if \/ else command to perform simple queries on all data types. The Microsoft Excel equivalent would be the “IF” feature. In R, the syntax is as follows:<\/p>\n
\r\nx <- 3                                # define x\r\n\r\nx <= 4                                # logical condition\r\n## [1] TRUE\r\n\r\nif (x <= 4) {\r\n  print("x is smaller than or equal to 4!")    \r\n} else {\r\n  print("x is larger than 4!")        \r\n}\r\n## [1] "x is smaller than or equal to 4!"\r\n<\/pre>\n
Explanation: An IF-command needs three parts: the keyword if()<\/span>, a condition that results in a single logical output x <= 4<\/span> and a block of code in curly braces {}<\/span>, which is executed if the expression is TRUE. So, if the condition is TRUE, the code will run in curly brackets after the IF-command. If the condition is FALSE, the code block after the else<\/span>is executed.
\nThe print()<\/span> function outputs the strings in the parentheses to the console window. Here we need the print()<\/span> function so that the output can be written out of the IF-function, similar to return<\/span>(see chapter 5<\/a>). A vectorized (and therefore more efficient) notation is the following:<\/p>\n
\r\nx <- c(3, 4, 5, 6, 7)                                \r\n\r\nifelse(x <=4, "x is smaller than or equal to 4!", "x is larger than 4!")\r\n## [1] "x is smaller than or equal to 4!" "x is smaller than or equal to 4!"\r\n## [3] "x is larger than 4!"              "x is larger than 4!"             \r\n## [5] "x is larger than 4!"\r\n<\/pre>\n
Using ifelse()<\/span>, the condition for each individual element is determined as a vector. This is helpful, e.g., if we want to categorize data!<\/p>\n
The FOR loop<\/strong><\/p>\n
Loops are incredibly useful when certain tasks need to be repeated very often in the script. A for loop is based on an iterable variable of defined length. But what does that mean? We define any variable, e.g., i<\/span>, with a start integer value, e.g, 1. We then increment this integer value until a second integer value, e.g. 8, is reached. This can be done via the sequence operator :<\/span>:<\/p>\n
\r\nfor (i in 1:8) {\r\n  print(i)\r\n}\r\n## [1] 1\r\n## [1] 2\r\n## [1] 3\r\n## [1] 4\r\n## [1] 5\r\n## [1] 6\r\n## [1] 7\r\n## [1] 8\r\n<\/pre>\n
The practical thing: The code in curly brackets is automatically executed once each time (eight times in total)! And we can meanwhile pick up the expression of our variable i<\/span>, in order to print it or index a vector with it and much more:<\/p>\n
\r\nv <- c(23, 54, 12, 59, 67, 45)    # create integer vector\r\n\r\nlength(v)                         # check length of vector\r\n## [1] 6\r\n  \r\nfor (i in 1:length(v)) {          # iterate length(v) times\r\n  print(v[i])\r\n}\r\n## [1] 23\r\n## [1] 54\r\n## [1] 12\r\n## [1] 59\r\n## [1] 67\r\n## [1] 45\r\n<\/pre>\n
It is also possible to set the variable \/ iterator equal to the elements of the vector instead of an integer value for indexing. The information in which run the loop is, however, is initially lost:<\/p>\n
\r\nv <- c("R", "is", "still", "fun")    \r\n\r\nfor (i in v) {\r\n  print(i)\r\n}\r\n## [1] "R"\r\n## [1] "is"\r\n## [1] "still"\r\n## [1] "fun"\r\n<\/pre>\n
Now, have a look at exercise V:<\/p>\n

\n