/* EXERCICE 1 */

/* question 1 */
data Mercury;
infile "Mercury.txt" expandtabs ;
 input id lake$ alkalinity ph calcium cholophyll avg_m nosamples minn maxx yrstd age_data ;
run;
/* question 2 */
proc print data=mercury;
run;    /* 53 obs */
/* question 3 */
data mercury_age;
  set mercury;
  if age_data=1;
  drop age_data;  /* 43 obs et 11 var */
  data no_age;
  set mercury;
  if age_data=0;
   drop age_data; /* 10 obs et 11 var */
  proc print data=mercury_age; run;
  proc print data=no_age; run;
  /* question 4 */
  proc means data=mercury_age;
  var alkalinity ph calcium cholophyll avg_m nosamples minn maxx yrstd; 
run; /* pas de données manquantes */
 proc corr data=mercury_age;
 var alkalinity ph avg_m yrstd; run; /* toutes les variables sont corrélées entre elles */
 /* question 5 */
 proc univariate data=mercury_age normal;
 var alkalinity yrstd; 
   histogram alkalinity yrstd/normal; 
  run; 
   /* question 6 */
  data n_mercury_age;
   set mercury_age;
   log_alkalinity=log(alkalinity);
   log_yrstd=log(yrstd);
   run;
   /* question 7 */
   proc univariate data=n_mercury_age normal;
 var log_alkalinity log_yrstd;   /* toutes les deux sont de loi Normale */
   histogram log_alkalinity log_yrstd/normal; 
  run; 
  /* question 8  */
proc means data=n_mercury_age;
  var log_alkalinity log_yrstd;
  run;   /* 2.99 et -0.82 */
  /* question 9 */
  data n2_mercury_age;
     set n_mercury_age;
	 diff=log_alkalinity-log_yrstd; 
	  proc univariate data=n2_mercury_age;
	    var diff;
		run;
		/* ou */
		proc ttest data=n2_mercury_age ;
		  var diff;
		  run;
		  /* pas la meme moyenne */
/* question 10 */
data n3_mercury_age;
   set mercury_age;
if calcium<25 then conc="faible";
if calcium>=25 & calcium<75 then conc="moyen";
if calcium>75 then conc="eleve";
proc print;
run;
/* question 11 */
Proc freq data=n3_mercury_age;
table conc /chisq;
run;
/* les trois proportions ne sont pas égales */
 

/* Exercice 2 */
proc sort data=mercury ;
	    by pH; run;
		axis1 value=(c=green) label=(c=green "alkalinity");
		axis2 value=(c=black) label=(c=black "pH");
	  proc gplot data=mercury;
	  plot alkalinity*pH / vaxis=axis1 haxis=axis2;
	  symbol1 interpol=j color=blue v=star;
	  title "alkalinity fonction du pH";
	  run;
quit;

/*  Exercice 3 ***/
	  /* question 1 */
	  option mprint mlogic symbolgen;
%macro calcul(tab,var,seuil);
data nouv ;
  set &tab ;
  if &var<&seuil;
  data _Null_;
    set nouv nobs=n;
  call symput ("nligne",n);
run;
%put "le nombre d observations du tableau &tab avec la variable &var inferieure à &seuil est egale à &nligne"; run;
%mend;

%calcul(mercury,ph,5) ; /* 7 observations */
%calcul(mercury,calcium,25) ;  /* 37 observations */