1          OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;
 55         
 56         /* MathReg3.sas  */
 57         %include '/folders/myfolders/441s16/Lecture/readexplor.sas';
 NOTE: Format YNFMT is already on the library WORK.FORMATS.
 NOTE: Format YNFMT has been output.
 NOTE: Format CRSFMT is already on the library WORK.FORMATS.
 NOTE: Format CRSFMT has been output.
 NOTE: Format NFMT is already on the library WORK.FORMATS.
 NOTE: Format NFMT has been output.
 
 NOTE: PROCEDURE FORMAT used (Total process time):
       real time           0.00 seconds
       cpu time            0.00 seconds
       
 
 174                 /* Creates data table explore */
 175        %include '/folders/myfolders/441s16/Lecture/readreplic.sas';
 
 NOTE: The infile '/folders/myfolders/exploremath.data.txt' is:
       Filename=/folders/myfolders/exploremath.data.txt,
       Owner Name=root,Group Name=vboxsf,
       Access Permission=-rwxrwx---,
       Last Modified=18Jan2016:17:34:49,
       File Size (bytes)=44583
 
 NOTE: 579 records were read from the infile '/folders/myfolders/exploremath.data.txt'.
       The minimum record length was 75.
       The maximum record length was 75.
 NOTE: Missing values were generated as a result of performing an operation on missing values.
       Each place is given by: (Number of times) at (Line):(Column).
       99 at 83:24    99 at 119:18   
 NOTE: The data set WORK.EXPLORE has 579 observations and 35 variables.
 NOTE: DATA statement used (Total process time):
       real time           0.02 seconds
       cpu time            0.01 seconds
       
 
 NOTE: Format YNFMT is already on the library WORK.FORMATS.
 NOTE: Format YNFMT has been output.
 NOTE: Format CRSFMT is already on the library WORK.FORMATS.
 NOTE: Format CRSFMT has been output.
 NOTE: Format NFMT is already on the library WORK.FORMATS.
 NOTE: Format NFMT has been output.
 
 NOTE: PROCEDURE FORMAT used (Total process time):
       real time           0.00 seconds
       cpu time            0.00 seconds
       
 
 288                 /* Creates data table replic  */
 289        title2 'Predict Grade for Replication Sample';
 290        
 291        /* Plan:
 292        
 293        1. Non-obvious findings from the exploration (based on Model I,
 294            which predicts grade from hsgpa hscalc hsengl totscore mtongue) were
 295              a. HS Engl neg
 296              b. mtongue neg
 297              c. totscore positive (diagnostic test matters)
 298           Test these on the replication with a Bonferroni correction for 3 tests.
 299           The other two results (HS GPA and HS Calculus) were obvious.
 300        
 301        2. See if prediction intervals work as advertised for Model H, which
 302           predicts grade from hsgpa hscalc hsengl totscore.
 303        
 304        3. Compare prediction of letter grade for the models with and without
 305           the diagnostic test.
 306        
 307        
 308        First, just illustrate use of different data tables in the same run.  */
 309        
 
 NOTE: The infile '/folders/myfolders/replicmath2.data.txt' is:
       Filename=/folders/myfolders/replicmath2.data.txt,
       Owner Name=root,Group Name=vboxsf,
       Access Permission=-rwxrwx---,
       Last Modified=30Jan2016:15:16:16,
       File Size (bytes)=38214
 
 NOTE: 579 records were read from the infile '/folders/myfolders/replicmath2.data.txt'.
       The minimum record length was 64.
       The maximum record length was 64.
 NOTE: Missing values were generated as a result of performing an operation on missing values.
       Each place is given by: (Number of times) at (Line):(Column).
       81 at 203:24   81 at 239:18   
 NOTE: The data set WORK.REPLIC has 579 observations and 35 variables.
 NOTE: DATA statement used (Total process time):
       real time           0.01 seconds
       cpu time            0.01 seconds
       
 310        proc freq data = explore;
 
 311             title3 'Exploratory Sample';
 312             tables outcome;
 
 NOTE: There were 579 observations read from the data set WORK.EXPLORE.
 NOTE: PROCEDURE FREQ used (Total process time):
       real time           0.04 seconds
       cpu time            0.05 seconds
       
 313        proc freq data = replic;
 
 314             title3 'Replication Sample';
 315             tables outcome;
 316        
 317        /* Now test the three findings: Point 1 above */
 318        
 
 NOTE: There were 579 observations read from the data set WORK.REPLIC.
 NOTE: PROCEDURE FREQ used (Total process time):
       real time           0.02 seconds
       cpu time            0.02 seconds
       
 319        proc reg data = replic  plots = none;
 
 320             title3 'Try to replicate HS Engl neg, mtongue neg, totscore pos';
 321             title4 'with a Bonferroni correction (check p < 0.05/3 = 0.01666667)';
 322             model grade = hsgpa hscalc hsengl totscore mtongue;
 323        
 324        /* Make combined data table, look at prediction intervals: Point 2 */
 325        
 
 NOTE: PROCEDURE REG used (Total process time):
       real time           0.07 seconds
       cpu time            0.06 seconds
       
 326        data predict;
 
 327             set explore replic;
 328             keeper =  grade+hsgpa+hscalc+hsengl+totscore;
 329             /* keeper will be missing if any of the vars are missing */
 330             if keeper ne .; /* Discards all other cases */
 331             grade2 = grade; /* Save value of grade for future use */
 332             if sample=2 then grade=. ;
 333             /* Response variable is now missing for replication sample.
 334                But it is preserved in grade2 */
 335        
 
 NOTE: Missing values were generated as a result of performing an operation on missing values.
       Each place is given by: (Number of times) at (Line):(Column).
       486 at 328:21   21 at 328:27    4 at 328:34     65 at 328:41    
 NOTE: There were 579 observations read from the data set WORK.EXPLORE.
 NOTE: There were 579 observations read from the data set WORK.REPLIC.
 NOTE: The data set WORK.PREDICT has 582 observations and 37 variables.
 NOTE: DATA statement used (Total process time):
       real time           0.01 seconds
       cpu time            0.01 seconds
       
 336        proc reg plots = none   data = predict;
 
 337                             /* Data table predict is the default anyway */
 338             title3 'Model H: hsgpa hscalc hsengl totscore: R-sq = 0.4532';
 339             model grade = hsgpa hscalc hsengl totscore;
 340             output out = predataH   predicted = Yhat
 341                                     L95       = lowpred
 342                                     U95       = hipred;
 343             /* Data table predataH has everything in predict plus
 344                Yhat and the lower and upper 95% predictoipn limits. */
 345        
 
 NOTE: The data set WORK.PREDATAH has 582 observations and 40 variables.
 NOTE: PROCEDURE REG used (Total process time):
       real time           0.06 seconds
       cpu time            0.06 seconds
       
 346        proc print;
 
 347             title3 'Look at predictions for the replication sample';
 348             var id sample grade2 Yhat lowpred hipred;
 349             where sample = 2;
 350        /* Should predicted marks be used to advise students? */
 351        
 352        /* Does 95 Percent Prediction Interval really contain 95 percent of grades?
 353           Recall that the data fail all tests for normality, and the prediction
 354           intervals are based on normal theory.  */
 355        
 
 NOTE: There were 293 observations read from the data set WORK.PREDATAH.
       WHERE sample=2;
 NOTE: PROCEDURE PRINT used (Total process time):
       real time           0.48 seconds
       cpu time            0.48 seconds
       
 356        data predictB;
 
 357             set predataH;
 358             if (lowpred < grade2 < hipred) then ininterval='Yes';
 359                else ininterval='No';
 360        
 
 NOTE: There were 582 observations read from the data set WORK.PREDATAH.
 NOTE: The data set WORK.PREDICTB has 582 observations and 41 variables.
 NOTE: DATA statement used (Total process time):
       real time           0.00 seconds
       cpu time            0.01 seconds
       
 361        proc freq;
 
 362             title3 'Does 95 Percent Prediction Interval Work?';
 363             tables sample * ininterval / nocol nopercent;
 364        
 365        /* Keep trying. Try to predict letter grade. */
 366        
 
 NOTE: There were 582 observations read from the data set WORK.PREDICTB.
 NOTE: PROCEDURE FREQ used (Total process time):
       real time           0.03 seconds
       cpu time            0.03 seconds
       
 367        data predictC;
 
 368             set predictB;
 369             if         80 <= grade2 <= 100 then lgrade = 'A';
 370                else if 70 <= grade2 <=  79 then lgrade = 'B';
 371                else if 60 <= grade2 <=  69 then lgrade = 'C';
 372                else if 50 <= grade2 <=  59 then lgrade = 'D';
 373                else if  0 <= grade2 <=  49 then lgrade = 'F';
 374             label lgrade = 'Letter Grade';
 375             pregrade = round(Yhat);
 376             if         80 <= pregrade <= 100 then prelgrade = 'A';
 377                else if 70 <= pregrade <=  79 then prelgrade = 'B';
 378                else if 60 <= pregrade <=  69 then prelgrade = 'C';
 379                else if 50 <= pregrade <=  59 then prelgrade = 'D';
 380                else if  0 <= pregrade <=  49 then prelgrade = 'F';
 381             label prelgrade = 'Predicted Letter Grade';
 382        
 
 NOTE: There were 582 observations read from the data set WORK.PREDICTB.
 NOTE: The data set WORK.PREDICTC has 582 observations and 44 variables.
 NOTE: DATA statement used (Total process time):
       real time           0.01 seconds
       cpu time            0.01 seconds
       
 383        proc freq;
 
 384             title3 'Accuracy of predicting Letter Grades From Model H';
 385             tables sample*prelgrade*lgrade / nocol nopercent;
 386             /* Will yield separate table for each sample. */
 387        
 388        
 389        /* Predict grade for a new student with hsgpa=80 hscalc=90 hsengl=70
 390           totscore=15. For just a prediction (no interval), proc glm is easier. */
 391        
 
 NOTE: There were 582 observations read from the data set WORK.PREDICTC.
 NOTE: PROCEDURE FREQ used (Total process time):
       real time           0.08 seconds
       cpu time            0.08 seconds
       
 392        proc glm data = explor;
 
 393             model grade = hsgpa hscalc hsengl totscore;
 394             estimate 'New Student' intercept 1  hsgpa 80  hscalc 90  hsengl 70
 395                                    totscore 15;
 396        
 397        /* Prediction for Y_{n+1} is the same as estimate of E[Y|X]. CI from proc glm
 398           is for E[Y|X]. PREDICTION interval for Y_{n+1} is wider. */
 399        
 
 NOTE: PROCEDURE GLM used (Total process time):
       real time           0.08 seconds
       cpu time            0.08 seconds
       
 400        data student;
 
 401             hsgpa=80; hscalc=90; hsengl=70; totscore=15; id = -27;
 402        
 
 NOTE: The data set WORK.STUDENT has 1 observations and 5 variables.
 NOTE: DATA statement used (Total process time):
       real time           0.00 seconds
       cpu time            0.02 seconds
       
 403        data together;
 
 404             set explore student;
 405             /* All variables not assigned will be missing for observation -27 */
 406        
 
 NOTE: There were 579 observations read from the data set WORK.EXPLORE.
 NOTE: There were 1 observations read from the data set WORK.STUDENT.
 NOTE: The data set WORK.TOGETHER has 580 observations and 35 variables.
 NOTE: DATA statement used (Total process time):
       real time           0.00 seconds
       cpu time            0.00 seconds
       
 407        proc reg plots = none;
 
 408             title3 'Model H: hsgpa hscalc hsengl totscore: R-sq = 0.4532';
 409             model grade = hsgpa hscalc hsengl totscore;
 410             output out = guess      predicted = PredictedY
 411                                     L95       = LowerLimit
 412                                     U95       = UpperLimit;
 413        
 
 NOTE: The data set WORK.GUESS has 580 observations and 38 variables.
 NOTE: PROCEDURE REG used (Total process time):
       real time           0.06 seconds
       cpu time            0.05 seconds
       
 414        data newguess;
 
 415             set guess;
 416             if id < 0; /* Discard all other cases */
 417        
 
 NOTE: There were 580 observations read from the data set WORK.GUESS.
 NOTE: The data set WORK.NEWGUESS has 1 observations and 38 variables.
 NOTE: DATA statement used (Total process time):
       real time           0.01 seconds
       cpu time            0.01 seconds
       
 418        proc print;
 
 419             title3 'hsgpa=80 hscalc=90 hsengl=70 totscore=15';
 420             var predictedY LowerLimit UpperLimit;
 421        
 422        
 423        
 424        
 425        
 426        
 427        
 428        
 429        
 430        
 431        OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;
 443