* STAT 3A03 Fall 2017; 
* SAS code for Assignment 3;

*Q1(a);
ods pdf file="A3Q1a.pdf";
PROC REG Data=S3A3.Races plots=none;
    Model Time=Distance Climb;
run;
ods pdf close;

*Q1(b);
PROC REG Data=S3A3.Races plots=none noprint;
    Model Time=Distance Climb;
    Plot Time*Pred.;
    Plot Student.*Pred.;
    Plot Student.*nqq.;
run;

*Q1(c);
PROC REG Data=S3A3.Races plot=none noprint;
    Model Time=Distance Climb;
    Plot CookD.*Obs.;
    Output Out=Races_out
        Predicted=Fitted
        Residual=Res_raw
        H=Leverage;
run;

PROC REG Data=S3A3.Races plot=none;
    model Time=Distance Climb;
    ods output anova=races_anova; /* save the ANOVA table */
run;

Data races_anova;
    set races_anova;
    If source='Error' then call symput ('races_sse', ss);
    If source='Model' then call symput ('races_p', df);
run;

Data Races_out;
    set Races_out;
    index=_N_;
    d=Res_raw/sqrt(&races_sse);
    potential=Leverage/(1-Leverage);
    residual_fun=(&races_p+1)*d**2/((1-Leverage)*(1-d**2));
    Hadi=potential+residual_fun;
run;

PROC GPLOT Data=Races_out;
    Plot Hadi*index;
    Plot Potential*Residual_fun;
run;

PROC GPLOT Data=Races_out;
   Plot Potential*index;
   Plot Residual_fun*index;
run;

*Q1(d);
Data Races_del7;
    Set S3A3.Races;
    If _N_=7 Then Delete;
run;

Data Races_del11;
    Set S3A3.Races;
    If _N_=11 Then Delete;
run;

Data Races_del18;
    Set S3A3.Races;
    If _N_=18 Then Delete;
run;

PROC REG Data=Races_del7 plots=none;
    model Time=Distance Climb;
run;

PROC REG Data=Races_del11 plots=none;
    model Time=Distance Climb;
run;

PROC REG Data=Races_del18 plots=none;
    model Time=Distance Climb;
run;
quit;

*Q2(a);
PROC REG Data=S3A3.Tab4_8 plots=none;
    Model Y=X1 X2 X3 X4 X5 X6;
    PLOT Y*Predicted.;
    PLOT Student.*Predicted.;
    PLOT Student.*nqq.;
run;
quit;

*Q2(b);
PROC REG Data=S3A3.Tab4_8 plots=none;
    Model Y=X1-X6;
    ods output anova=Tab4_8_anova; /* save the ANOVA table */
    output out=Tab4_8out
        residual=res_raw
        student=res_stud
        CookD=C
        DFFITS=DFITS
        H=Leverage;
run;

Data Tab4_8_anova;
    set Tab4_8_anova;
    If source='Error' then call symput ('Q2_sse', ss);
    If source='Model' then call symput ('Q2_p', df);
run;

Data Tab4_8out;
    set Tab4_8out;
    Obs=_N_;
    d=res_raw/sqrt(&Q2_sse);
    potential=Leverage/(1-Leverage);
    residual_fun=(&Q2_p+1)*d**2/((1-Leverage)*(1-d**2));
    Hadi=potential+residual_fun;
run;

*Q2(c);
PROC GPLOT Data=Tab4_8out;
   PLOT res_stud*Obs;
   PLOT C*Obs;
   PLOT DFITS*Obs;
   PLOT Hadi*Obs;
   PLOT potential*residual_fun;
run;
quit;

*Q3(a);
ODS PDF FILE="A3Q3a.pdf";
PROC REG Data=S3A3.Tab4_8 plots=none;
    MODEL Y=X1-X4 / partial;
run;
ODS PDF CLOSE;

*Q3(b);
ODS PDF FILE="A3Q3b.pdf";
PROC REG Data=S3A3.Tab4_8 plots=none;
   MODEL Y=X1-X5 / partial;
run;
ODS PDF CLOSE;

*Q3(c);
ODS PDF FILE="A3Q3c.pdf";
PROC REG Data=S3A3.Tab4_8 plots=none;
   MODEL Y=X1-X4 X6 / partial;
run;
ODS PDF CLOSE;

*Q3(d);
ODS PDF FILE="A3Q3d.pdf";
PROC REG Data=S3A3.Tab4_8 plots=none;
   MODEL Y=X1-X3;
run;
ODS PDF CLOSE;