Question

我有一个数据集，其中包含三列：station_code，dest_code，fare。基本上，station_code和dest_cde内的数据相同，而票价是前往特定车站所需的费用。

station_code dest_code fare
station1     station1  0
station1     station2  4.6
station1     station3  10
station1     station4  10
station1     station5  12.3
station1     station6  12.3
station1     station7  12.3
station1     station8  12.3
station1     station9  14.7 
station1     station10 14.7
           .
           .
           .
station1     station91 27.5
station2     station1  4.6
station2     station2  0
station2     station3  10
station2     station4  10
station2     station5  12.3
station2     station6  12.3
station2     station7  12.3
station2     station8  12.3
station2     station9  14.7 
station2     station10 14.7
          .
          .
          .

耕种站91

所以我的问题是如何使用数组技术创建看起来像这样的查找表。

fee  1    2    3    4   ...
1    0    4.6  10   10
2    4.6  0    10   10
3    10   10   0    4.6
4    10   10   4.6  0
5    12.3 12.3 4.6  4.6
...  ...  ...  ...  ...

如您所见，行和列中的索引实际上都代表站点名称，例如row1 = station1，column1 = station1，column2 = station2。

Answer 1

问题是如此混乱，我想再次澄清一下我的表和查找表。

对于第一张图片，它是原始表格 enter image description here

enter image description here 第二张图片是我想要的查找表

Answer 2

您可以通过PROC TRANSPOSE获得类似的表格，并提前对数据进行排序。

 proc sort data=have ; 
 by station_code dest_code fare;
 run;

 proc transpose data=have out=want;
 by station_code;
 id dest_code;
 var fare;
 run;

Answer 3

您将fare描述为二维数组。如何“加载”阵列取决于您计划如何使用“查找”。

假设：

在一个数据集中的票价：station_code，dest_code，fare
- 站代码值的字面意思是station1…station91
初始点：personid，step_num，station_code
您要计算每个人的总票价

示例：

data totals(keep=personid totalfare);
  * load the station fares into temporary array for use as lookup table;
  array fares(91,91) _temporary_;
  do until (lastfare);
    set fares end=lastfare;
    _from = input(substr(station_code,8),best.);  * parse the number out of code;
    _dest = input(substr(dest_code,8),best.);
    fares(_from,_dest) = fare;
  end;

  * compute each persons total fare;
  do until (endtrips);
    totalfare = 0;
    _from = 0;
    do until (last.personid);
      set trips end=endtrips;
      by personid step_num;

      _dest = input(substr(station_code,8),best.);
      if _from and _dest then totalfare + fares(_from,_dest);
      _from = _dest;
    end;
    output;
  end;

  stop;
run;

如果工作站代码值实际上不是不是一个可以解析1…91的值，则不能使用数组 -相反，应使用具有两个值的键的哈希对象作为查找对象。

data totals (keep=personid totalfare);
  * load the station fares into hash for use as lookup table;
  if 0 then set fares; * prep pdv;
  declare fares hash(dataset:'fares');
  fares.defineKey('dest_code', 'station_code'); * reversed keys make it easier to traverse trips;
  fares.defineData('fare');
  fares.defineDone();  * automatically reads dataset:fares and fills hash entries;

  * compute each persons total fare;
  do until (endtrips);
    totalfare = 0;
    dest_code = '';
    do until (last.personid);
      set trips end=endtrips;                  * read in the station for a persons step;
      by personid step_num;

      if fares.find()=0 then do;  * 0 return code means variable fare has the value for the fare from station_code to dest_code;
        totalfare + fare;
      end;

      * prepare for next leg of journey, this is what is meant by easier to traverse;
      dest_code = station_code;
    end;
    output;
  end;

  stop;
run;

如何使用数组创建查找表

3 个答案: