* project1.sas Christine Iodice 061209 Model Hunt;
* Course project team: 
    Christine Iodice, Prasad Satich, Saranne Warner, Daniel Brockman ;
* Course X446 SAS Data Analysis. Jian-min Liu, Instructor. ;

options ls=200 nocenter mlogic mprint symbolgen;

*libname source 'C:\Documents and Settings\Christine\Desktop\SAS_Project';
libname source 'C:\b\SAS-X446\prj';

/* View Source Data*/
Title 'Original source data';

Proc print data=source.air;
run;

Proc contents data=source.air;
run;


title;

/* Create temporary dataset w/ labels*/
Data source.air1 (Drop=AVS CSM);
	set source.air;
	*if CPM = 4.737 then delete;
	Seats = SPA*1000;
	Avgfilled = ALF*seats; *create variable for filled seats;
	AvgEmpty = (1-ALF)*seats; *create variable for empty seats;
	Fill_to_Empty = Avgfilled / AvgEmpty;
	LCPM = Log(CPM);
	LUTL = Log(UTL);
	LASL = Log(ASL);
	LSPA = Log(SPA);
	LALF = log(ALF);
    SUTL = UTL**2;
	SASL = ASL**2;
	SSPA = SPA**2;
	SALF = ALF**2;

	label CPM = 'CPM: Cost per Passenger Mile (cents)'
          UTL = 'UTL: Avg Hrs per Day in Use'
		  ASL = 'ASL: Avg Length of nonstop legs (1000 miles)'
		  SPA = 'SPA:Avg Nbr of Seats per Aircraft (per 1000 seats)'
		  Seats = 'Seats:Actual Seats on aircraft'
		  ALF = 'ALF: Avg Load Factor (% occupied)'
		  Type = 'Type: Range (short or long)'
		  AvgEmpty = 'Average Nbr of Empty Seats'
		  Avgfilled = 'Average Nbr of Filled Seats'
		  Fill_to_Empty = 'Ratio filled to Empty';
run;

/*Very Cool Scatterplot matrix*/

%include 'C:\Documents and Settings\Christine\Desktop\SAS_Project\scattmat.sas';
%Scattmat(source.air1, CPM UTL ASL SPA ALF Fill_to_Empty)




Proc contents data=source.air1;
run;

Proc Sort data=source.air1;
	by type ASL CPM;
run;

Title 'Listing of data = air1';
Proc Print data=source.air1 width=min label;
Title 'sorted by Type, ASL, CMP';
run;


Data source.AirShort source.airLong;
	set source.air1;
	if type = 0 then output source.airshort;
	If type = 1 then output source.airlong;
run;


Proc Reg data=source.air1;
	Model CPM = UTL ASL SPA ALF Type /selection=stepwise p r influence vif; 
Plot student.*(UTL ASL SPA ALF);
Plot Student.*predicted. cookd.*obs.;
Plot Npp.*Residual.;
run;
Quit; 

/* Log tranformation give Best R-Squared which is .59 */

Proc Reg data=source.air1;
	Model CPM = LUTL LASL LSPA LALF Type /selection=stepwise p r influence vif; 
Plot student.*(UTL ASL SPA ALF);
Plot Student.*predicted. cookd.*obs.;
Plot Npp.*Residual.;
run;
Quit; 

Proc Reg data=source.air1;
	Model CPM = UTL SUTL ASL SASL SPA ALF SALF Type /selection=stepwise p r influence vif; 
run;
Quit;