【SAS NOTES】实际分析应用

data guanhui.visit_8_12;/*导入客户访问情况*/
    infile 'E:\****.txt' dlm='|' firstobs=2;
    input user :$11. serv_number :$11. log_time :$16. log_IP :$15. content :$100.;
    log_date=scan(log_time,1,' ');
        log_year=scan(log_date,1,'-');
        log_month=scan(log_date,2,'-');
        log_day=scan(log_date,3,'-');
    log_clock=scan(log_time,2,' ');
        log_hour=scan(log_clock,1,':');
        log_minute=scan(log_clock,2,':');
    visit_time=DHMS(mdy(log_month,log_day,log_year),log_hour,log_minute,0);
run;
proc print data=guanhui.visit_8_12 (firstobs=1 obs=12);
    format visit_time datetime18.;
run;
data guanhui.buydetail;/*交易明细汇总*/
    infile 'E:\****.txt' dlm='|' firstobs=2;
    input serv_number :$11. pay_way :$50. fee pay_date :$19. pay_month :$10. goods_name :$100. goods_detail :$100.;
    buy_date=scan(pay_date,1,' ');
    buy_clock=scan(pay_date,2,' ');
        buy_year=scan(buy_date,1,'-');
        buy_month=scan(buy_date,2,'-');
        buy_day=scan(buy_date,3,'-');
        buy_hour=scan(buy_clock,1,':');
        buy_minute=scan(buy_clock,2,':');
        buy_second=scan(buy_clock,3,':');
    buy_time=DHMS(mdy(buy_month,buy_day,buy_year),buy_hour,buy_minute,buy_second);
run;
proc print data=guanhui.buydetail (firstobs=1 obs=12);
    format buy_time datetime18.;
run;
data guanhui.existuser;/*存量客户明细*/
    infile 'E:\****.txt' dlm='|' firstobs=2;
    input user :$20. serv_number :$11. logon_date :$15.;
        logon_year=scan(logon_date,1,'-');
        logon_month=scan(logon_date,2,'-');
        logon_day=scan(logon_date,3,'-');
        logon_time=mdy(logon_month,logon_day,logon_year);
run;
proc print data=guanhui.existuser (firstobs=1 obs=10);
    format logon_time mmddyy8.;
run;
data guanhui.activeuser;/*2012年活跃用户*/
    infile 'E:\*****.txt' dlm='09'x firstobs=2;
    input user :$15. serv_number :$11. lastlog_time :$20.;
    lastlog_date=scan(lastlog_time,1,' ');
        lastlog_year=scan(lastlog_date,1,'-');
        lastlog_month=scan(lastlog_date,2,'-');
        lastlog_day=scan(lastlog_date,3,'-');
    lastlog_clock=scan(lastlog_time,2,' ');
        lastlog_hour=scan(lastlog_clock,1,':');
        lastlog_minute=scan(lastlog_clock,2,':');
    lastlog_t=DHMS(mdy(lastlog_month,lastlog_day,lastlog_year),lastlog_hour,lastlog_minute,0);
run;
proc print data=guanhui.activeuser (firstobs=1 obs=10);
    format lastlog_t datetime18.;
run;
/*以上部分为数据导入部分，下面为数据分析部分*/
data work.buedetail_same;
    set guanhui.buydetail;
    drop buy_date buy_clock buy_year buy_month buy_day buy_hour buy_minute buy_second pay_date pay_month;
    format buy_time datetime18.;
    month=MONTH(datepart(buy_time));
    day=DAY(datepart(buy_time));
    hour=HOUR(buy_time);
    weekday=WEEKDAY(datepart(buy_time));
    quter=QTR(datepart(buy_time));
run;
proc print data=work.buedetail_same (firstobs=1 obs=100);
run;
proc tabulate data=work.buedetail_same out=guanhui.buydetail_evedetailfee;
    class month weekday hour goods_name;
    var fee;
    table month,(max min mean sum median)*fee;
    table weekday,(max min mean sum median)*fee;
    table hour,(max min mean sum median)*fee;
    table goods_name,(max min mean sum median)*fee;
run;
proc print data=guanhui.buydetail_evedetailfee;
run;

数据导入主要为 : $ scan mdy dhms 等函数及字符处理运用。

在检查中print可以用(firstobs=1 obs=10)来显示部分数据。

在分析处理部分中：

1、用data set 来从源数据中取数以及生成新变量来进行分析处理。

2、此次常规分析汇总功能用proc tabulate过程。

3、在常规分析中对日期的处理分析经常遇到，在源数据中形成一个完整日期格式的变量，在分析中注意对该变量的函数处理需要相应格式的支持。所以datepart()取出date变量中日期部分很重要，否则会出错。