Question

我正在尝试使用proc sql折叠我的数据。但是，我注意到当我试图折叠我的数据时，我丢失了一堆我想要保留的变量。我试图根据变量MRN（数字）折叠我的数据。我想要保留的其他变量是CITY和SITE（这些是字符值），并且这些变量对于每个唯一的MRN都是常量，因此折叠它们应该没问题。

以下是我正在使用的代码

proc sql;
    create table collapsed_data  as 
        select distinct mrn,
            sum(msk_tx_yes) as msk_tx_yes,
            sum(msk_cancel_tx_yes) as msk_cancel_tx_yes,
            sum(msk_ca_yes) as msk_ca_yes, 
            sum(msk_cancel_ca_yes) as msk_cancel_ca_yes, 
            sum(msk_dc_yes) as msk_dc_yes,
            sum(conc_psych_tx_yes) as conc_psych_tx_yes,
            sum(conc_psych_ca_yes) as conc_psych_ca_yes,
            sum (conc_psych_dc_yes) as conc_psych_dc_yes,
            sum (conc_yes) as conc_yes,
            sum (psych_yes) as psych_yes,
            sum (foot_prog) as foot_prog,
            sum (hand_prog) as hand_prog,
            sum (surg_prog) as surg_prog,
            sum (sx_yes) as sx_yes
        from temp_collapsed_data
        group by mrn;
quit;

我不确定如何一起使用SELECT和DISTINCT功能。

我想也许我可以在SELECT之后添加变量CITY和STATE，同时保持DISTINCT，但它确实可以工作。

我希望能够将CITY和STATE保留在新表中以及我正在制作的新求和变量中。如何在不将CITY和STATE转换为虚拟编码变量的情况下实现这一目标？如果可能，我想将它们保留为字符值。

任何人都知道如何实现这一目标？

Answer 1

你的代码已经正确了。只需将变量添加到select语句中即可。

proc sql;
    create table collapsed_data  as 
        select distinct mrn, city, site,
            sum(msk_tx_yes) as msk_tx_yes,
            sum(msk_cancel_tx_yes) as msk_cancel_tx_yes,
            sum(msk_ca_yes) as msk_ca_yes, 
            sum(msk_cancel_ca_yes) as msk_cancel_ca_yes, 
            sum(msk_dc_yes) as msk_dc_yes,
            sum(conc_psych_tx_yes) as conc_psych_tx_yes,
            sum(conc_psych_ca_yes) as conc_psych_ca_yes,
            sum (conc_psych_dc_yes) as conc_psych_dc_yes,
            sum (conc_yes) as conc_yes,
            sum (psych_yes) as psych_yes,
            sum (foot_prog) as foot_prog,
            sum (hand_prog) as hand_prog,
            sum (surg_prog) as surg_prog,
            sum (sx_yes) as sx_yes
        from temp_collapsed_data
        group by mrn;
quit;

distinct语句将导致没有两行具有相同的信息。

如何在保留其他变量的同时折叠数据？

1 个答案: