/* */

OPTIONS pagesize=40 linesize =80 NOCENTER ; run;

DATA storms;

*Mac path...;

INFILE "Macintosh HD:Courses:613(StatSoftware):storms.csv"
  delimiter = "," missover ;

*Windows path...;

*INFILE "C:/.../storms.csv"
  delimiter = "," missover ;

INPUT YEAR
   N_STORMS
   N_CAT
   N_HS_HIT
   N_HR_1_9
   N_HR_1_5
   N_W1
   N_W2
   N_W3
   N_W4
   N_W5
   N_H_HIT
   N_HR
   N_TS
   N_SS
;

label N_STORMS  = "No. storms in database";
label N_CAT     = "No. with entry in <<Hit US>> category";
label N_HS_HIT  = "No. Made landfall over US as trop_storm /hurricane";
label N_HR_1_9  = "No. where category > 0";
label N_HR_1_5  = "No. that hit U.S. as hurricane coded 1-5";
label N_W1      = "No. that hit US with strength 1";
label N_W2      = "No. that hit US with strength 2";
label N_W3      = "No. that hit US with strength 3";
label N_W4      = "No. that hit US with strength 4";
label N_W5      = "No. that hit US with strength 5";
label N_H_HIT   = "No. that hit US with strength  > 0";
label N_HR      = "No. where Max intensity = HR = hurricane";
label N_TS      = "No. where Tp = TS = tropical storm";
lable N_SS      = "No. where Tp = SS = subtropical storm";

decade = 10*INT(YEAR/10);   /* use built-in fn. to compute decade         */
                            /* INT( ) returns integer portion of argument */

IF YEAR < 1900 then 19thcent = 1 ;  /* e.g. create /alter a variable  */
ELSE                19thcent = 0 ;  /* e.g. create /alter a variable  */

IF YEAR ne . ;  /* put obsn. into dataset only if year is not missing */

          /* NOTE: 2 uses of <<IF>> ... */

          /* 1. IF(condition); obsn. included in dataset only if condn. is TRUE */

          /* 2. IF(condition) THEN ...  further statements(s)           */
          /*                            e.g. create /alter a variable   */


RUN;

Proc CONTENTS DATA= storms; RUN;

TITLE "PROC MEANS  -- CLASS statement subdivides observations ";

Proc MEANS data = storms N MIN MAX MEAN SUM MAXDEC=1;
 CLASS decade;
 VAR   N_HR_1_5;
RUN;

TITLE "Most PROCs can direct a copy of output to new SAS dataset ...  ";

Proc MEANS data = storms  N MEAN SUM MAXDEC=1;
 CLASS decade;
 VAR   N_HR_1_5;
 OUTPUT OUT = stats N = No_years  MEAN = ave_yr SUM = total;

RUN;

TITLE "Print out the observations in new dataset  ";

PROC PRINT DATA = STATS; RUN;

TITLE "PLOT the mean no. per decade versus decade in new dataset  ";

PROC PLOT DATA = STATS;
  PLOT Ave_yr * decade;
RUN;

TITLE "exclude the nineteenth century observations from PLOT ";

PROC PLOT DATA = STATS;
  PLOT Ave_yr * decade;
  WHERE (decade >= 1900 );

RUN;

TITLE "Plot number (rather than mean) per decade, and exclude 1990s ";

PROC PLOT DATA = STATS;
  PLOT total * decade;
  WHERE (decade >= 1900 AND decade <=1980 );

RUN;

TITLE "Variability in numbers over decades 1900s to 1980s ";

PROC MEANS DATA = STATS MAXDEC=1;
  Var total ;
  WHERE (decade >= 1900 AND decade <=1980 );

RUN;


TITLE "direct statistics for several variables to new SAS dataset ...  ";

Proc MEANS data = storms  N MEAN SUM NOPRINT;
 CLASS decade;
 VAR   N_HR_1_5 N_W1 - N_W5;
 OUTPUT OUT = stats   MEAN = ave_15 ave_1 - ave_5 SUM = tot15 tot1 - tot5;

PROC PRINT DATA = STATS ROUND;
RUN;