SPSS-等级和分区

Question

在SPSS Statistics语法文件中，我希望创建一个变量，该变量根据所需的分区列（例如，等效于SQL“在Oracle SQL开发人员中排名（按column_a按b划分列））”来计算排名。 / p>

请参见示例：

没有任何过滤器的初始数据：

应用get_rank后的最终输出：

Answer 1

To create a rank variable as described, first sort your data and then use the LAG function.

SORT CASES BY column_a column_b .
IF ($CASENUM=1) rank=1 .
IF ($CASENUM>1 AND column_a~=LAG(column_a)) rank=1 .
IF ($CASENUM>1 AND column_a=LAG(column_a)) rank=LAG(rank) + 1 .
EXE .

LAG will look at the value of column_a for the prior case. In the syntax above it checks whether the value in column_a is different from that of the prior case.

If it has, then it will set the rank to 1. If it hasn't, then it will add 1 to the rank of the prior case. Just make sure your data is properly sorted first.

From there, if you want to look only at records that are rank=1, you can either use FILTER BY or SELECT IF to do that.

Answer 2

如果确实只需要过滤key=1的密钥，则可以使用以下方法：

SORT CASES BY column_a column_b .
match files /file=* /by column_a /first=key1.

现在，变量key1对于column_a类别的第一次出现将具有值1，您可以将其用于filter或select。

对于完整排名变量，您可以使用此变量（甚至不需要先排序）：

RANK VARIABLES=b (A) BY a /RANK  /TIES=MEAN.