It is most common to read data in with white

1
Chapter 12
2

• It is most common to read data in with white
• space between observations and one record
• per line.
• data new
• input x y z
• datalines
• 2 no 4
• 8 yes 6
• 2 no 6

3
Reading data from a file

• data new
• infile etest.dat
• input x y z
• run
• The data is in a separate text file on drive E.

4
Comma-delimited data

• data new
• infile etest.dat dlm,
• input x y z
• run
• The data
• 2, no, 4
• 8, yes, 6
• 2, no, 6

5
Delimiter Sensitive Data(DSD)

• data new
• infile etest.dat dsd
• input x y z
• run
• The data
• 2,, 4
• 8, yes, 6
• 2, no, 6

• When there are 2 adjacent commas, the observation
  is treated as missing.
• When there are quotes around a value for a
  character variable, they are removed.

6
Informats

• We saw informats used with dates in Chapter 4.
• Informats can be used to extend the length of a
  long value.
• They can also be used to combine values.

7
Eight-character limit

• Suppose we run the program below. See the
• output on the next page.
• data new
• input x y z
• datalines
• 4 tennessee 20
• 14 georgia 3
• 22 mississippi 2
• proc print
• run

8

• Notice that SAS cut off tennessee and mississippi
• when the 8 character limit was reached.

9

• data new
• informat y 13.
• input x y z
• datalines
• 4 tennessee 20
• 14 georgia 3
• 22 mississippi 2
• proc print
• run

10

• data new
• input x y 13. z
• datalines
• 4 tennessee 20
• 14 georgia 3
• 22 mississippi 2
• proc print
• run

11

• Suppose that you want to combine the first and
• last name into one single variable.
• data new
• input age name 30. gpa
• datalines
• 23 jennifer smith 3.1
• 26 bob simpson 2.4
• 33 greg davis 3.6

12

• One caution about this.
• In the list of data, there are at least 2 spaces
  between
• where the name ends and the gpa begins.
• If there was only 1 space between then we would
  have
• the output on the following page.

13

• This is all the output that was printed before
  SAS
• complained because it was expecting a numeric
  for
• X but found bob which is clearly not.

14
Column Input

• data new
• input x 1-2 y 3-5 z 6-8
• datalines
• 12113455
• 98832211
• 87733101

15
Formatted Input

• data new
• input _at_1 x 2.
• _at_3 y 3.
• _at_6 z 3.
• datalines
• 12113455
• 98832211
• 87733101

16

• data new
• input _at_6 id 4.
• _at_32 average 4.
• datalines
• 960112103 jane smith 88.3
• 960252886 george washington 78.7
• 960312557 greg davis 92.1

17

• Sometimes one record takes up more than one
• line.
• data new
• input 1 _at_1 ID 1 means line 1
• _at_11 Name 30.
• 2 _at_ 11 GPA 2 means line 2
• datalines
• 960112103 jane smith
• 88.3
• 960252886 george washington
• 78.7
• 960312557 greg davis
• 92.1

18
(No Transcript)
19

• Variables can be read in any order.
• Columns of data may be read more than once.
• data new
• input x 1-3 y 2-5 z 1-9
• datalines
• 181822330
• 001451923

20
Informat Lists

• data course
• input _at_1 quiz1 2.
• _at_4 quiz2 2.
• _at_7 quiz3 2.
• _at_10 exam1 3.
• _at_14 exam2 3.
• datalines
• 10 14 15 88 93
• 16 12 17 82 94
• 19 20 13 81 90

21

• We can shorten the previous data step with
• data course
• input _at_1 (quiz1-quiz3)(2.)
• _at_10 (exam1-exam2)(3.)
• datalines
• 10 14 15 88 93
• 16 12 17 82 94
• 19 20 13 81 90

22

• A more compact way of
• doing this is
• data course
• input _at_1 (quiz1-quiz3)(2. 4)
• _at_4 (hwork1-hwork2)(2. 4)
• datalines
• 10 14 15 11 12
• 16 12 17 12 18
• 19 20 13 14 20

• Suppose we have
• data course
• input _at_1 quiz1 2.
• _at_4 hwork1 2.
• _at_7 quiz2 2.
• _at_10 hwork1 2.
• _at_13 quiz3 2.
• datalines
• 10 14 15 11 12
• 16 12 17 12 18
• 19 20 13 14 20

23
(No Transcript)
24
Holding the line

• data new
• input x _at__at_
• datalines
• 3 10 11 15
• 22 21 4 6

25

• data new
• input x y _at__at_
• datalines
• 3 10 11 15
• 22 21 4 6

26
Suppressing Error Messages

• Suppose that you receive a dataset and the
• person that created it used NA for missing
• observations rather than a period.
• Lets suppose were reading in a column of
• peoples ages. When SAS comes across this
• NA, it will complain.

27

• data new
• input x ??
• datalines
• 33
• 41
• 55
• 23
• NA
• 41
• 22