您好,我正在为一个暑期学校做项目,教书的人喜欢逐行工作。我认为将其转换为矩阵并与数组一起使用将是有益的。
当前数据集如下:
data longdata;
input var1 var2 Value;
datalines;
rowone colone 1
rowone coltwo 1
rowone colthree 2
rowone colfour 3
rowtwo colone 1
rowtwo coltwo 1
rowtwo colthree 1
rowtwo colfour 2
rowthree colone 3
rowthree coltwo 3
rowthree colthree 3
rowthree colfour 4
rowfour colone 1
rowfour coltwo 3
rowfour colthree 3
rowfour colfour 3
;
run;
我想将以上内容转换为此,最好通过by语句:
data matrixdata;
input var1-var4;
datalines;
colone coltwo colthree colfour
rowone 1 1 2 2
rowtwo 1 1 1 2
rowthree 3 3 3 4
rowfour 1 3 3 3
;
run;
答案 0 :(得分:4)
Proc transpose
非常适合将分类数据转换为其他布局形式。
proc transpose data=have out=want;
by var1 notsorted;
id var2;
var value;
run;
尽管看起来像数组,但数据不是隐式数组。要在单个隐式数据步骤迭代期间成为4x4数组,传入数据集实际上将需要每行16个变量,例如v1-v16
,其值为1 1 2 3 1 1 1 2 3 3 3 4 1 3 3 3
。每行将对应一个数组实例。
data want2;
retain v1-v16;
array v (16) v1-v16;
set have end=end;
v(_n_) = value;
if end;
keep v:;
run;
data _null_;
set want2;
array m(4,4) v1-v16;
do row = 1 to dim(m,1);
do col = 1 to dim(m,2);
put row= col= m(row,col)=;
end;
end;
run;