/* module 5: problems 5.1, 5.4 */
* for skewed data use percentiles;
* for symmetric data use mean and sd;

/* 5.1 */
filename utility 'Utility.dat';
data one;
  infile utility;
  input date $ 1-6 telcost 9-15 fuelcost 17-22 emcost 25-29;
  totcost = telcost + fuelcost + emcost;
   label telcost = Telephone Costs;
   label fuelcost = Fuel Costs;
   label emcost = Electricity Costs;
   label totcost = Total Costs;
run;
proc univariate plot;
  var telcost fuelcost;
  id date;
   title "Utility Costs: Telephone and Fuel";
run;
proc univariate plot;
  var emcost totcost;
  id date;
   title "Utility Costs: Electricity and Total";
run;
* telcost is positively skewed with a skewness of .991;
*   the minimum is 39.39 (Jan 92);
*   Q1 is 51.33;
*   the median is 69.02;
*   Q3 is 91.50;
*   the maximum is 145.17 (Jun 90);
*   no outliers;
* fuelcost is positively skewed with a skewness of .733;
*   the minimum is 16.21 (Jul 92);
*   Q1 is 23.72;
*   the median is 48.67;
*   Q3 is 105.94;
*   the maximum is 157.71 (Jan 91);
*   no outliers;
* emcost is positively skewed with a skewness of .903;
*   the minimum is 18.69 (Jun 92);
*   Q1 is 26.58;
*   the median is 30.07;
*   Q3 is 36.88;
*   the maximum is 56.61 (Dec 88, an outlier);
*   no outliers;
* totcost is positively skewed with a skewness of .597;
*   Max         317.41  (Sep 92);
*   Q3          204.72  ;
*   Median      166.04  ;
*   Q1          120.70  ;
*   Min         91.40   (Jan 91);
*   no outliers;

/* 5.4 */
filename china 'China#1.dat';
data two;
  infile china;
  input year 1-4 gross 6-10 exports 12-16 imports 18-22;
  balance = exports - imports;
   label
    year = Year
    gross = "Gross Trade (US$ billion)"
    exports = "Exports (US$ billion)"
    imports = "Imports (US$ billion)"
    balance = "Trade Balance (US$ billion)"
   ;
run;
proc univariate plot;
  var imports exports balance;
  id year;
   title Chinese Trade;
run;
* imports is positively skewed with a skewness of 1.254;
*   Max         39.85 ;
*   Q3          18.53 ;
*   Median      2.86  ;
*   Q1          1.89  ;
*   Min         1.17  ;
*   no outliers;
* exports is positively skewed with a skewness of 1.343;
*   Max         43.44 ;
*   Q3          20.89 ;
*   Median      3.44  ;
*   Q1          1.98  ;
*   Min         1.41  ;
*   no outliers;
* balance is negatively skewed with a skewness of -1.645;
*   Max         4.34  ;
*   Q3          0.43  ;
*   Median      0.14  ;
*   Q1         -0.23  ;
*   Min        -8.41  ;
*   low outliers in 78, 80, 79, 86, and 85;
*   high outliers in 81, 83, 89, and 82;