960.390 - Introduction to Computers for Statistics

960.390-01, Fall 1999, M 7,8 (6:10-9:00pm)

Meeting dates: 10/25, 11/1, 11/8, 11/15

---

| Syllabus | Class 1 | Class 2 | Class 3 | Class 4 | Class 5 | Home | Email

SAS FUNCTIONS:

SAS have many build-in functions for working with data values. They include numeric functions, character functions, probability and distribution functions, among others.

SAS Numeric Functions:

A numeric function provides a numerical computation on one or more numerical variables. See page 35 of the textbook for a list and description of some of the numeric functions

Example:

 
OPTIONS LINESIZE=65 NODATE; 
DATA ONE; 
 INPUT X @@; 
 Y = ABS(X); 
 Z = LOG(Y); 
CARDS; 
1.4 -1.5 0.35 0.66 -5 
; 
RUN; 
PROC PRINT; 
RUN; 
 
  The SAS System 1
 
 OBS X Y Z
 
 1 1.40 1.40 0.33647
 2 -1.50 1.50 0.40547
 3 0.35 0.35 -1.04982
  4 0.66 0.66 -0.41552
 5 -5.00 5.00 1.60944

SAS Character Functions:

A character function operates on one or a few character strings. Page 36 of the textbook lists some of the character functions.

Example:

 
OPTIONS LINESIZE=65 NODATE; 
DATA ONE; 
 INPUT @1 NAMES $CHAR15.; 
 R_ALIGN = RIGHT(NAMES); 
 L_ALIGN = LEFT(NAMES); 
CARDS; 
 ABLERT 
 ALEX 
NACY 
 EMMANUAL 
; 
RUN; 
PROC PRINT; 
RUN; 
 
 The SAS System  1
 
 OBS NAMES R_ALIGN L_ALIGN
 
 1 ABLERT ABLERT ABLERT 
 2 ALEX ALEX ALEX 
 3 NACY NACY NACY 
 4  EMMANUAL EMMANUAL EMMANUAL

SAS Probability and Distributional Functions:

· For a given distribution and a specified value in the sample space, a probability function returns the corresponding cumulative distribution function (probability) value.

--- See page 36 of the textbook for a list and description of the probability functions.

· For a given distribution (with specified parameter(s)) and a seed, a SAS distributional function generates a random observation from such a distribution.

--- See page 61 for more a list of some distributional functions and a description of the random observation that is generated.  

Example:

 
OPTIONS LINESIZE=65 NODATE; 
DATA ONE; 
 INPUT X @@; 
 N = 6; 
 P = 0.3; 
 CDF = PROBBNML(P,N,X); 
CARDS; 
0 1 2 3 4 5 6 
; 
RUN; 
PROC PRINT; 
RUN; 
DATA TWO; 
 Y = RANNOR(3333); 
 OUTPUT; 
RUN; 
PROC PRINT; 
 VAR Y; 
RUN; 
 
 The SAS System 1
 
 OBS X N P CDF
 
 1 0 6 0.3 0.11765
 2 1 6 0.3 0.42018
 3 2 6 0.3 0.74431
 4 3 6 0.3 0.92953
 5 4 6 0.3 0.98907
 6 5 6 0.3 0.99927
 7 6 6 0.3 1.00000
^L The SAS System 2
 
 OBS Y
 
 1 -0.41520

Do Loops and Generate Multiple Random Observations in SAS:

DO Loops:

Do loops are used when a statement or set of statements needs to be repeated many times.

General form of a do loop is

 
 DO index_variable = beginning number TO ending number;
  Statements;
 OUTPUT; /* You need this line when a DO loop is used in a DATA step, 
 otherwise these elements don't get added to the data set */ 
 END;
 

Example:

 
OPTIONS LINESIZE=65 NODATE;
DATA MYDATA;
 DO BASE = 0 TO 12;
 DEG = BASE * 30;
 RAD = (DEG * 3.14159) / 180;
 COSX = COS(RAD);
 SINX = SIN(RAD);
 TANX = TAN(RAD);
 OUTPUT;
 END;
RUN;
PROC PRINT;
RUN;

 
 The SAS System 1
 
 OBS BASE DEG RAD COSX SINX  TANX
 
 1 0 0 0.00000 1.00000 0.00000 0.00
 2 1 30 0.52360 0.86603 0.50000 0.58
 3 2 60 1.04720 0.50000 0.86602 1.73
 4 3 90 1.57080 0.00000 1.00000  753696.00
 5 4 120 2.09439 -0.50000 0.86603 -1.73
 6 5 150 2.61799 -0.86602 0.50000 -0.58
 7 6 180 3.14159 -1.00000 0.00000 -0.00
 8 7 210 3.66519 -0.86603 -0.50000  0.58
 9 8 240 4.18879 -0.50000 -0.86602 1.73
 10 9 270 4.71239 -0.00000 -1.00000 251232.00
 11 10 300 5.23598 0.50000 -0.86603 -1.73
 12 11 330 5.75958 0.86602 -0.50000  -0.58
 13 12 360 6.28318 1.00000 -0.00001 -0.00

The DO construction can also be added without an index, to extend the functionality of IF THEN ELSE constructions. Like this:

 
DATA MYDATA;
DO AGES = 10 TO 18;
 IF (AGES < 13) THEN DO
 TYPE = "PRE-TEEN";
 SCHOOL = "GRADE";
 END;
 ELSE DO
 TYPE = "TEEN";
 SCHOOL = "MID/HIGH";
 END;
 OUTPUT;
END;
RUN;
PROC PRINT;
RUN;
 
 
 OBS AGES TYPE SCHOOL
 
  1 10 PRE-TEEN GRADE
 2 11 PRE-TEEN GRADE
 3 12 PRE-TEEN GRADE
 4 13 TEEN MID/H
 5 14 TEEN MID/H
 6 15 TEEN MID/H
 7 16 TEEN MID/H
  8 17 TEEN MID/H
 9 18 TEEN MID/H

Use DO Loops and Distributional Functions to Generate Multiple Random Observations:

The following is an example to generate 10 random normal distribution (mean = 3, sd = 1.5), 10 exponential random variables (mean = 7), and 10 random numbers from a uniform distribution on (1,5).

 
OPTIONS LINESIZE=65 NODATE;
DATA ONE;
 DO INDX = 1 TO 10;
 XNORMAL = (RANNOR(3333) * 1.5) + 3;
 YEXP = RANEXP(44444)*7;
 ZUNI = RANUNI(1234567) * (5 - 1) + 1;
 OUTPUT;
 END;
RUN;
PROC PRINT;
RUN;

The SAS System 16

 
 The SAS System 1
 
 OBS INDX XNORMAL YEXP ZUNI
 
 1 1 2.37720 3.40962 3.30274
 2 2 1.98452 2.01766 1.51405
 3 3 3.27308 1.66074 1.24179
 4 4 4.90929 1.91629 3.18993
 5 5 1.55773  4.09273 4.79824
 6 6 2.96389 2.89952 2.29184
 7 7 2.46344 0.98380 3.74177
 8 8 2.00966 3.24449 3.03825
 9 9 3.05269 7.59491 3.57499
 10  10 1.28174 1.06131 1.95390

DISCRIPTIVE STATISTICS II

PROC FREQ Procedure:

PROC FREQ is used to generate a table or tables of frequency counts from data that are in categories. To generate a "one-way" table, simply use one variable. A one-way table summarizes all the values of the variable, how many variables have each value, and the percent and cumulative percent for each value. To generate a "two-way" table, use two variables. A two-way table contains cell frequencies, cell percent of total, cell percent of row total, and cell percent of column total.

The usage of PROC FREQ is:

 
 PROC FREQ DATA=DATASET;
 BY VARIABLE(S);
 TABLES V1 / options;
 TABLES V1*V2 / options;
 RUN;

There are substantial numbers of options, which can be found on Page 54:

Example:

 
OPTIONS LINESIZE=78 NODATE;
DATA MYDATA;
 DO I = 1 TO 500;
 VALP = RANPOI(3024,500);
 VALN = RANNOR(3024);
 VALU = (RANUNI(3024) * 100);
 IF (VALU > 80) THEN TYP = "GREAT";
 ELSE IF (VALU > 60) THEN TYP = "OK";
 ELSE TYP = "BAD";
 IF (VALN > 0) THEN SIGN = "POS";
 ELSE SIGN = "NEG"; 
 IF (VALP < 480) THEN PVAL = 480;
 ELSE IF (VALP < 520) THEN PVAL = 520;
 ELSE PVAL = 560; 
 OUTPUT;
 END;
RUN;
PROC FREQ DATA=MYDATA;
 TABLES PVAL;
 TABLES TYP*SIGN/CHISQ;
RUN;
 
 The SAS System 1
 
 Cumulative Cumulative
 PVAL Frequency Percent Frequency Percent
 --------------------------------------------------
 480 96 19.2 96 19.2
 520 317 63.4 413 82.6
 560 87 17.4 500 100.0
 
 
 
 
 TABLE OF TYP BY SIGN
 
 TYP SIGN
 
 Frequency|
 Percent |
 Row Pct |
 Col Pct |NEG |POS | Total
 ---------+--------+--------+
 BAD | 141 | 164 | 305
 | 28.20 | 32.80 | 61.00
 | 46.23 | 53.77 |
  | 59.49 | 62.36 |
 ---------+--------+--------+
 GREAT | 43 | 49 | 92
 | 8.60 | 9.80 | 18.40
 | 46.74 | 53.26 |
 | 18.14 | 18.63 |
 ---------+--------+--------+
 OK | 53 | 50 | 103
 | 10.60 | 10.00 | 20.60
 | 51.46 | 48.54 |
 | 22.36 | 19.01 |
 ---------+--------+--------+
 Total 237 263 500
 47.40 52.60 100.00
 
^L  The SAS System 2
 
 STATISTICS FOR TABLE OF TYP BY SIGN
 
 Statistic DF Value Prob
 ------------------------------------------------------
  Chi-Square 2 0.863 0.649
 Likelihood Ratio Chi-Square 2 0.862 0.650
 Mantel-Haenszel Chi-Square 1 0.736 0.391
 Phi Coefficient 0.042
  Contingency Coefficient 0.042
 Cramer's V 0.042
 
 Sample Size = 500
 

Also Study Example 7.1 on page 54 and Example 19.1 on page 141.