libname ref "/folders/myfolders";

/* Importing excel file */
proc import datafile ="/folders/myfolders/DataSet1.xlsx"
out= ref.dataset1
DBMS=XLSX replace;
sheet='Employee-Attrition';
getnames = yes;
run;

/* Importing csv file */
proc import datafile ="/folders/myfolders/CSVData.csv"
out= ref.csvdata
DBMS=csv replace;
getnames = yes;
run;


/* 1. High paying specialization in terms of daily rate? */
proc means data=ref.dataset1 sum;
var DailyRate;
class education;
run;
No.3 i.e. Bachelor

/* 2. High paying specialization by degree */


proc means data=ref.dataset1 sum;
var DailyRate;
class  EducationField ;
run;
/* Lifesciences */

proc means data=ref.dataset1 sum;
var DailyRate;
class EducationField education ;
run;
/* Bachelor degree in life sciences */

/*Frequency Distribution */
proc format;
value agefrmt
0 - <25='Less than 25'
25 - 34='25 -34'
35 - 41='35 -41'
42 - 50='42 -50'
51 - high='51 and above';
run;

/*Which age group employees has attrited most */
proc freq data = ref.dataset1;
tables age * Attrition / nopercent nocol nocum norow;
format age agefrmt.;
run;
/* 25 -34  */

Which age group has least traveled?
proc freq data=ref.dataset1;
tables age*BusinessTravel/ nopercent nocol nocum norow;
format age agefrmt.;
run;
/* less than 25 */

/* People who made frequent switches in the past generally leave an */
/* organization early  */
proc freq data=ref.dataset1;
tables age*NumCompaniesWorked/ norow nocol nopercent nocum;
format age agefrmt.;
run; 

proc freq data = ref.dataset1;
tables age * Attrition / nopercent nocol nocum norow;
format age agefrmt.;
run;
/* 25 -34  */


/* Which department has highest level of job satisfaction? */
proc format;
value jsfrmt
1 ='Low'
2 ='Medium'
3 ='High'
4 ='Very High';
run;

proc freq data=ref.dataset1;
tables Department*JobSatisfaction/ norow nocum nocol;
format JobSatisfaction jsfrmt.;
run;
/* Research & Development */