SAS Lecture 3

1
SAS Lecture 3

• Computing normal probabilities in SAS
• PROC CORR
• PROC GPLOT

2
Example Data on the time between machine
failures were collected during a study on machine
performance that involved 39 similar machines.
From the data we compute, the sample mean
23.35hours and the sample standard deviation
1.67h.

• What is the percentage of machines that failed
  before 20 hours?
• What is the percentage of machines that failed
  after 24 hours?
• What is the percentage of machines with failure
  time between 20 and 22 hours?
• How short should the failure time be for a
  machine to be in the bottom 10 ?

3
Computing the normal probabilities using SAS

• data normal
• input x _at__at_
• mean23.35
• stdev1.67
• z(x-mean)/stdev
• pnprobnorm(z)
• invpn1-probnorm(z)
• label pn"the normal dist. func. at x"
• label z"the standardized value"
• label invpn"the area to the right of x"
• datalines
• 20 24
• run
• proc print label
• run

4
SAS output

• the the normal
  the area
• standardized dist.
  to the
• Obs x mean stdev value func. at x
  right of x
• 20 23.35 1.67 -2.00599 0.02243
  0.97757
• 24 23.35 1.67 0.38922 0.65144
  0.34856

Answer to question 1
Answer to question 2
5
Computing the normal percentiles using SAS

• data percentile
• input p _at__at_
• mean23.35
• stdev1.67
• ipprobit(p)
• ipmeanstdevip
• label ip"the p-th percentile"
• datalines
• 0.1
• proc print label
• run

6
SAS output

• the p-th
• Obs p mean stdev percentile
• 1 0.1 23.35 1.67 21.2098

Answer to question 4
7
SAS procedures for scatter plots and correlation

• PROC CORR
• The CORR procedure is a statistical procedure for
  numeric random variables that computes Pearson
  correlation coefficients and some descriptive
  statistics. The correlation statistics include
• PROC CORR DATA dataset-name
• BY ltDESCENDINGgt variable-1 ltvariable-ngt
• VAR variable(s)
• WITH variable(s)

8

• Data one
• input time line step device
• linetline/1000
• datalines
• 0.0893 266 2 1
• 0.0386 120 1 1
• 0.0988 245 2 1
• 0.026 102 1 2
• 0.041 307 2 2
• 0.0196 143 1 2
• proc corr
• var time line step
• run

9

• The CORR Procedure
• 3 Variables time line step
• Simple Statistics
• Variable N Mean Std Dev Sum
  Minimum Maximum
• time 6 0.05222 0.03349 0.31330
  0.01960 0.09880
• line 6 197.16667 86.06374 1183
  102.0000 307.00000
• step 6 1.50000 0.54772 9.0
  1.0000 2.00000
• Pearson Correlation Coefficients, N
  6
• Prob gt r under H0 Rho0
• time line step
• time 1.00000 0.61490
  0.78996
• 0.1939 0.0615
• line 0.61490 1.00000
  0.96099
• 0.1939 0.0023
• step 0.78996 0.96099
  1.00000
• 0.0615 0.0023

10

• proc corr
• var time
• with line step
• run
• Produces the correlations between time and line,
  and time and step only.

The CORR Procedure
2 With Variables line step 1
Variables time Simple
Statistics Variable N Mean Std Dev
Sum Minimum Maximum line 6 197.16667
86.06374 1183 102.0000 307.000 step
6 1.50000 0.54772 9.00 1.0000
2.0000 time 6 0.05222 0.03349
0.313 0.0196 0.0988 Pearson
Correlation Coefficients, N 6
Prob gt r under H0 Rho0
time line 0.61490
0.1939 step
0.78996 0.0615
11

• proc sort
• by device
• proc corr
• by device
• var time line
• run
• The BY statement specifies the variable that the
  procedure uses to form BY groups. The data need
  to be sorted first by the BY variable. This
  procedure will compute the correlation between
  time and line for the two groups of data defined
  by the variable device

12

• ----------------------- device1
  --------------------------
• The CORR Procedure
• 2 Variables time line
• Pearson Correlation Coefficients, N 3
• Prob gt r under H0 Rho0
• time line
• time 1.00000 0.96086
• 0.1787
• line 0.96086 1.00000
• 0.1787
• ----------------------- device2
  -----------------------------
• The CORR Procedure
• 2 Variables time line
• Pearson Correlation Coefficients, N 3
• Prob gt r under H0 Rho0
• time line

13
PROC GPLOT

• SYMBOL is a global option that controls the plot
  display.
• SYMBOLlt1 2 399gt
• ltCOLORsymbol-colorgt ? control the point color
• ltVALUEspecial-symbol text-string NONEgt
  ?change the plotting symbol
• ltINTERPOLJOINgt ? join the points with a line
• ltINTERPOL R ? draw the regression line through
  the cloud of points
• PROC GPLOT creates a scatter plot for two
  variables.
• PROC GPLOT ltDATAinput-data-setgt
• PLOT xvaryvarltzvargt
• BY variables ? construct a different plot for
  each group defined by the BY variables.

14
Possible value for symbol

• You can use any of the following symbols
• Asterisk Diamond
• Plus Square
• Circle Dot
• Point Star
• Club Heart
• Spade Triangle

15

• symbol interpolr valuedot colorred
• proc gplot
• plot bratelgnp
• run

16

• proc sort
• by cg
• run
• proc gplot
• plot brategnpcg
• run

Add a categorical variable
17
symbol1 interpolnone valueE colorred symbol2
interpolnone valueS colorblack symbol3
interpolnone valueG colorgreen symbol4
interpolnone valueM colorblue symbol5
interpolnone valueA colormagenta symbol6
interpolnone valueF colorbrown proc
gplot plot brategnpcg run
18

• To change the scales and tick marks on the axes
• data kilowatt
• input kwh ac dryer _at__at_
• datalines
• 35 1.5 1 63 4.5 2 66 5.0 2 17 2.0 0 94 8.5 3 79
  6.0 3 93
• 13.5 1 66 8.0 1 94 12.5 1 82 7.5 2 78 6.5 3 65
  8.0 1 77
• 7.5 2 75 8.0 2 62 7.5 1 85 12.0 1 43 6.0 0 57 2.5
  3 33
• 5.0 0 65 7.5 1 33 6.0 0
• proc gplot data kilowatt
• axis1 0 to 20 by 5
• axis2 0 to 100 by 20
• symbol valueplus colorblue
• plot kwhac / haxisaxis1 vaxisaxis2
• title "Plot of KWH against AC"
• run