* Christine Iodice 061211 Results of myTop3.sas ;

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

* 061211 db: 1. Minor operational adjustments
*            2. addition of model im1 
*            3. removed stepwise selection from m5xASL
*            4. restated m5xASL to omit ASL and TYPE, correcting error
*;


options ls=80 nocenter mlogic mprint symbolgen;

libname source 'C:\Documents and Settings\Christine\Desktop\SAS_Project';
libname source 'c:\b\SAS-X446\prj' ;     * location of data;
%let orig=air_jianmin_23Nov06 ;          * orig data file ;
%let clean=air_clean_pct ;               * orig data excluding extreme values ;
%let clean=air_clean_str ;
%let extremes=air_ex_pct ;               * excluded values data file ;
%let extremes=air_ex_str ;

     ********************************************;
     * select input data set to use for models   ;
     *                                           ;
            %let inuse=&orig;     
     *                                           ;
     ********************************************;


* end parameters ------------------------------- ;



title;

/* Create temporary dataset w/ labels*/
Data source.air1 (Drop=AVS CSM);
	set source.&inuse;

	*/below are potential outliers to delete.;
	*if CPM = 4.737 then delete; *identified as outlier through plots;
    *if ALF = .287 then delete;  *identified as outlier through plots;

	*if CPM = 2.341 then delete; *identified as outlier through stars method;
    *if CPM = 3.306 then delete; *identified as outlier through stars method;
	
    * not creating bucket variable for plane size;
	if 0.0 <= SPA <= 0.1 then psize=1;
    if 0.1  < SPA <= 0.2 then psize=2;
    if 0.2  < SPA <= 0.3 then psize=3;
	if 0.3  < SPA <= 0.4 then psize=4;	
	*now creating bucket variables for ASL;
	if  0   <= ASL <= 500 then pdistance = 1;
	if  500  < ASL <= 1000 then pdistance = 2;
	if  1000 < ASL  <= 1500 then pdistance = 3;
	if  1500 < ASL  <= 2000 then pdistance = 4;
	if  2000 < ASL  <= 2500 then pdistance = 5;
	if  2500 < ASL  <= 3000 then pdistance = 6;
	if  3000 < ASL  <= 3500 then pdistance = 7;
	if  3500 < ASL  <= 4000 then pdistance = 8;

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);
	LFilled = log(avgfilled);
	Lempty = log (avgempty);
    Lpsize = log(psize);
	Lpdistance = log(pdistance);
	SUTL = UTL**2;
	SASL = ASL**2;
	SSPA = SPA**2;
	SALF = ALF**2;
	Sfilled = Avgfilled**2;
	Sempty = Avgempty**2;

	* db: Vars for IM1 ---------------------- ;
	 ISPA  = 1/SPA ;
     IALF  = 1/ALF ;
	* --------------------------------------- ;
 
	
	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'
          Psize = 'Plane Size: 1 is lowest';
run;

/*
Proc print data=source.air1;
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 print data=source.airshort;
run;

Proc print data=source.airlong;
run;
*/

Ods pdf file = 'C:\Documents and Settings\Christine\Desktop\SAS_Project\Top3Results.pdf';

/*Below is my Choice 1 */
Proc Reg data=source.air1;
Air1B: Model CPM =  ASL LUTL LFilled Lempty / vif influence p r;
Title 'Air1: CPM = ASL LUTL Lfilled Lempty';
Plot Student.*predicted. cookd.*obs.;
Plot Npp.*Residual.;

run;
quit;

/*Below is my choice #2 */
/*this is old model 5 -- but changed to SLE = .05, ASL drops out --*/
Proc reg data=source.air1;
	/* db m5xASL: model CPM = UTL ASL SPA ALF Type  / */
 	m5xASL: model CPM = UTL SPA ALF /
		/* db Selection = stepwise SLE=.15 SLS = .05 */ vif influence p r;
	/* db Title " Air1: Stepwise Regression of original variables"; */
		Title "m5xASL: model5, without var ASL";
run;
quit;

/* Short - Long modelling */

/*this is the best for airshort */
/* db Proc reg data = source.airshort;  */
Proc reg data = source.air1;  
Long1: Model CPM = Lfilled Lempty UTL / stb vif influence p r;
Title 'Airshort: CPM = Lfilled Lempty UTL';
run;
quit;

/*This is the best for airlong */ 

/* db Proc Reg data=source.airlong; */
Proc Reg data=source.air1;
Long3: Model CPM =Lfilled lpsize/ stb vif influence p r;
Title 'Airlong:CPM = Lfilled lpsize';
run;
Quit;



/* db im1 */
Proc Reg data=source.air1;
 im1:  model CPM=ISPA IALF / stb vif influence p r;
Title 'im1:  model CPM=ISPA IALF';
run;
Quit;

ods pdf close;