* h-01 Daniel Brockman 061211 Influence Analysis -- Air;
* Course project team: 
    Christine Iodice, Prasad Satich, Saranne Warner, Daniel Brockman ;
* Course X446 SAS Data Analysis. Jian-min Liu, Instructor. ;

Title 'Influence analysis';

Title2 'Program h-01.sas';
options ls=80;

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

* 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 extremes=air_ex_pct ;               * excluded values data file ;


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


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

title;

Title1 "Original Data" ;
* sort the data before further use ;
data orig4 ;
  set source.&orig ;
run;
proc sort data=orig4;
  by descending CPM ;
run;
proc print data=orig4 ;
run;


Title1 "Excluding extreme observations";
* sort the data before further use ;
data clean4 ;
  set source.&clean ;
run;
proc sort data=clean4;
  by descending CPM ;
run;
proc print data=clean4 ;
run;

Title1 ;

* The data we are running with ;
* sort the data before further use ;
data inuse3 ;
  set source.&inuse ;
run;
proc sort data=inuse3;
  by descending CPM ;
run;
  



* cook up some variable transformations;
data air2;
  set inuse3 ;
  
	Seats = SPA*1000;
	AvgFilled = ALF*seats; *create variable for filled seats;
	AvgEmpty = (1-ALF)*seats; *create variable for empty seats;

  IAVGF = 1/AvgFilled ;
  ISPA  = 1/SPA ;
  IALF  = 1/ALF ;
  IASL  = 1/ASL ;

  * BUCKETIZATION ;
  * now 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;
  
	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;


run;

  


Title1 'Isolating Observations of Influence';

proc reg data=air2 outest=airregest;

  Long3: Model CPM =Lfilled lpsize / stb vif influence p r ; 
  Long2: Model CPM = Lfilled Lempty / stb vif influence p r ;
  im1:  model CPM=ISPA IALF / stb vif influence p r ;

  
              

run;


*----------------------------------------;