SAS 数据分析实例之数据描述、预处理和抽样

坚强别无选择

1 描述统计
1.1 描述表属性

1. ods noproctitle;
   
2. ods select attributes variables enginehost directory;
   
3. 
   
4. proc datasets lib=SASHELP;
   
5.         contents data=SASHELP.CARS order=collate
   
6.                 out=WORK.TableAttributes (label="Contents Details for SASHELP.CARS");
   
7. quit;
   
8. 
   
9. proc print;
   
10. run;

复制代码

1.2 描述数据特征
1.2.1 分析分类变量

1. title "分类变量的频数";
   
2. 
   
3. proc freq data=SASHELP.CARS;
   
4.         tables Make Model Type Origin DriveTrain / plots=(freqplot) missing;
   
5. run;

复制代码

1.2.2 分析数值变量

1. title "数值变量的描述性统计量";
   
2. 
   
3. proc means data=SASHELP.CARS n nmiss min mean median max std;
   
4.         var MSRP Invoice EngineSize Cylinders Horsepower MPG_City MPG_Highway Weight
   
5.                 Wheelbase Length;
   
6. run;
   
7. 
   
8. title;
   
9. 
   
10. proc univariate data=SASHELP.CARS noprint;
    
11.         histogram MSRP Invoice EngineSize Cylinders Horsepower MPG_City MPG_Highway
    
12.                 Weight Wheelbase Length;
    
13. run;

复制代码

1.3 描述缺失数据

1. ods noproctitle;
   
2. 
   
3. proc format;
   
4.         value _nmissprint low-high="非缺失";
   
5.         value $_cmissprint " "=" " other="非缺失";
   
6. run;
   
7. 
   
8. proc freq data=SASHELP.CARS;
   
9.         title3 "缺失数据频数";
   
10.         title4 h=2 "图例: .、A、B，其他 = 缺失";
    
11.         format MSRP Invoice EngineSize Cylinders Horsepower MPG_City MPG_Highway
    
12.                 Weight Wheelbase Length _nmissprint.;
    
13.         format Make Model Type Origin DriveTrain $_cmissprint.;
    
14.         tables Make Model Type Origin DriveTrain MSRP Invoice EngineSize Cylinders
    
15.                 Horsepower MPG_City MPG_Highway Weight Wheelbase Length / missing nocum;
    
16. run;
    
17. 
    
18. proc freq data=SASHELP.CARS noprint;
    
19.         table Make * Model * Type * Origin * DriveTrain * MSRP * Invoice * EngineSize
    
20.                 * Cylinders * Horsepower * MPG_City * MPG_Highway * Weight * Wheelbase *
    
21.                 Length / missing out=Work._MissingData_;
    
22.         format MSRP Invoice EngineSize Cylinders Horsepower MPG_City MPG_Highway
    
23.                 Weight Wheelbase Length _nmissprint.;
    
24.         format Make Model Type Origin DriveTrain $_cmissprint.;
    
25. run;
    
26. 
    
27. proc print data=Work._MissingData_ noobs label;
    
28.         title3 "跨变量的缺失数据模式";
    
29.         title4 h=2 "图例: .、A、B，其他 = 缺失";
    
30.         format MSRP Invoice EngineSize Cylinders Horsepower MPG_City MPG_Highway
    
31.                 Weight Wheelbase Length _nmissprint.;
    
32.         format Make Model Type Origin DriveTrain $_cmissprint.;
    
33.         label count="频数" percent="百分比";
    
34. run;
    
35. 
    
36. title3;
    
37. 
    
38. proc delete data=Work._MissingData_;
    
39. run;

复制代码

2 数据预处理
2.1 列出数据

1. title1 "列出数据 - SASHELP.BASEBALL";
   
2. 
   
3. proc sort data=SASHELP.BASEBALL out=WORK.SORTTEMP;
   
4.         by Team;
   
5. run;
   
6. 
   
7. proc print data=WORK.SORTTEMP label n;
   
8.         var Position;
   
9.         by Team;
   
10.         sum Salary;
    
11. run;
    
12. 
    
13. proc delete data=work.SORTTEMP;
    
14. run;
    
15. 
    
16. title1;

复制代码

2.2 过滤数据

1. proc sql noprint;
   
2.         create table WORK.filter as select * from SASHELP.BASEBALL where(Salary LT 100
   
3.                 AND Position EQ "CF");
   
4. quit;

复制代码

2.3 排序数据

1. proc sort data=SASHELP.BASEBALL out=WORK.sortDS noequals;
   
2.         by descending Salary;
   
3. run;

复制代码

2.4 排名数据

1. proc rank data=SASHELP.BASEBALL descending out=WORK.Rank;
   
2.         var Salary;
   
3.         ranks rank_Salary;
   
4. run;

复制代码

2.5 转换数据

1. data WORK.transform;
   
2.         set SASHELP.BASEBALL;
   
3.         log_Salary=log(Salary);
   
4.         inv_CrHits=1 / CrHits;
   
5. run;

复制代码

2.6 标准化数据

1. proc stdize data=SASHELP.BASEBALL method=std nomiss out=WORK.Stdize oprefix
   
2.                 sprefix=Standardized_;
   
3.         var Salary nHits nAtBat;
   
4. run;

复制代码

2.7 重置码值

1. data WORK.recodedValues;
   
2.         set SASHELP.BASEBALL;
   
3. 
   
4.         select (Salary);
   
5.                 when (100) _recodeVar_=99999;
   
6.                 when (200) _recodeVar_=88888;
   
7.                 otherwise _recodeVar_=Salary;
   
8.         end;
   
9. run;

复制代码

2.8 重置码范围

1. data WORK.recodedRanges;
   
2.         set SASHELP.BASEBALL;
   
3. 
   
4.         select;
   
5.                 when (-1 <=Salary <=100) _recodeVar_=100;
   
6.                 otherwise _recodeVar_=Salary;
   
7.         end;
   
8. run;

复制代码

3 随机抽样
3.1 简单随机抽样
3.1.1 无放回不重复抽样

1. proc surveyselect data=SASHELP.BASEBALL out=WORK.RandomSample
   
2.                 method=srs samprate=0.3 seed=2019;
   
3.         strata Position / alloc=prop;
   
4. run;

复制代码

3.1.2 有放回重复抽样

1. proc surveyselect data=SASHELP.BASEBALL out=WORK.RandomSample
   
2.         outhits method=urs sampsize=30 seed=2019;
   
3.         strata Position / alloc=prop;
   
4. run;

复制代码

3.2 分层随机抽样
3.2.1 无放回不重复抽样

1. proc sort data=SASHELP.BASEBALL out=WORK.SORTTempTableSorted;
   
2.         by Position;
   
3. run;
   
4. 
   
5. proc surveyselect data=WORK.SORTTempTableSorted out=WORK.RandomSample
   
6.                 method=srs samprate=0.3 seed=2019;
   
7.         strata Position / alloc=prop;
   
8. run;

复制代码

3.2.2 有放回重复抽样

1. proc sort data=SASHELP.BASEBALL out=WORK.SORTTempTableSorted;
   
2.         by Position;
   
3. run;
   
4. 
   
5. proc surveyselect data=WORK.SORTTempTableSorted out=WORK.RandomSample
   
6.         outhits method=urs sampsize=30 seed=2019;
   
7.         strata Position / alloc=prop;
   
8. run;

复制代码

karstification

看一看，谢谢