1. THIS IS DATAPLOT PROGRAM FILE QUINLAN.DP 2. NOISE/SHRINKAGE OF SPEEDOMETER CASING 3. JIM QUINLAN 4. TAGUCHI ANALYSIS EXAMPLE 5. 2**(15-11) ANALYSIS 6. INPUT DATA FILE--QUINLAN.DAT 7. OUTPUT FILES--DPPL1F.DAT (A POSTSCRIPT FILE CONTAINING ALL PLOT OUTPUT) 8. OUT1. (YATES ANALYSIS OUTPUT) 9. OUT2A. (PREDICTED VALUES AND RESIDUALS FROM FINAL MODEL) 10. OUT2B. 11. OUT3A. (SORTED Y AND CARRY-ALONG X'S) 12. OUTBB. 13. 14. -----START POINT---------- 15. 16DIMENSION 40 VARIABLES 17DEVICE 2 POSTSCRIPT 18. 19. STEP 1--READ IN THE DATA 20. 21SKIP 50 22READ QUINLAN.DAT X1 TO X15 Y1 TO Y4 23LOOP FOR K = 1 1 15 24LET X^K = -1 SUBSET X^K 1 25LET X^K = +1 SUBSET X^K 2 26END LOOP 27LET Y = (Y1+Y2+Y3+Y4)/4 28DELETE Y1 Y2 Y3 Y4 29. 30. STEP 2--GENERATE DEX SCATTER PLOT & DEX MEAN PLOT 31. 32MULTIPLOT 2 1 33LABEL SIZE 4 34X3LABEL AUTOMATIC 35CHAR X; LINES BLANK 36DEX SCATTER PLOT Y X1 TO X15 37CHAR BLANK; LINES SOLID 38DEX MEAN PLOT Y X1 TO X15 39LINES SOLID ALL; LINES BLANK BLANK 40. 41. STEP 3--GENERATE MATRIX OF MEAN PLOTS (USE DEXSTAT2.DP MACRO) 42. 43ER 44MULTIPLOT OFF 45X3LABEL 46LET NUMFAC = 15 47LET DP = 1 48LET STRING STAT = MEAN 49YLIMITS 0 .6 50LET STRING S0 = SHRINKAGE 51LET STRING S1 = X1 52LET STRING S2 = X2 53LET STRING S3 = X3 54LET STRING S4 = X4 55LET STRING S5 = X5 56LET STRING S6 = X6 57LET STRING S7 = X7 58LET STRING S8 = X8 59LET STRING S9 = X9 60LET STRING S10 = X10 61LET STRING S11 = X11 62LET STRING S12 = X12 63LET STRING S13 = X13 64LET STRING S14 = X14 65LET STRING S15 = X15 66CALL DEXSTAT2.DP 67YLIMITS 68. 69. STEP 4--GENERATE BLOCK PLOTS 70. 71. NONE DONE HERE 72. 73. STEP 5--DO YATES ANALYSIS (DATA MUST BE IN YATES ORDER) 74. NOTE--COEFFICIENTS STORED OUT TO FILE DPST1F.DAT 75. 76ER 77YATES Y 78CAPTURE OUT1. 79YATES Y 80END CAPTURE 81. 82. STEP 6--GENERATE NORMAL AND HALFNORMAL PROBABILITY PLOTS ON THE EFFECTS 83. NOTE--MUST READ COEFFICIENTS IN FROM FILE DPST1F.DAT 84. 85SKIP 0 86READ DPST1F.DAT TAG COEF T RESSD1 RESSD2 87WRITE TAG COEF T RESSD1 RESSD2 88CHAR X; LINES BLANK 89LABEL SIZE 4 90X3LABEL AUTOMATIC 91MULTIPLOT 2 2 92NORMAL PROBABILITY PLOT COEF 93LET ABSCOEF = ABS(COEF) 94HALFNORMAL PROBABILITY PLOT ABSCOEF 95. 96. STEP 7--GENERATE PREDICTED VALUES AND RESIDUALS FOR PARTIAL MODEL 97. 98LET MU = MEAN Y 99LET B5 = COEF(1) 100LET B5 = -B5 101LET B7 = COEF(2) 102LET RESSD = RESSD2(2) 103LET PRED = MU + 0.5*(B5*X5 + B7*X7) 104LET RES = Y-PRED 105ER 106SET WRITE DECIMALS 3 107WRITE X1 X2 X3 X4 X5 X6 X7 X8 X9 X1 108WRITE X11 X12 X13 X14 X15 Y PRED RES 109WRITE OUT2A. X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 110WRITE OUT2B. X11 X12 X13 X14 X15 Y PRED RES 111. 112. STEP 8--GENERATE NORMAL PROBABILITY PLOT OF RESIDUALS 113. 114MULTIPLOT 2 2 115X1LABEL PRED = ^MU + 0.5*(^B5*X5 + ^B7*X7) 116X2LABEL RESIDUAL STANDARD DEVIATION = ^RESSD 117LABEL SIZE 3 118. PLOT RES RUNSEQ 119PLOT RES 120NORMAL PROBABILITY PLOT RES 121X1LABEL 122X2LABEL 123LABEL SIZE 2 124MULTIPLOT OFF 125. 126. STEP 9--GENERATE CONTOUR PLOT FOR THE 2 MOST IMPORTANT FACTORS 127. (USE MACRO DEXCONT.DP) 128. 129LET YCONT = DATA 0 .1 .2 .3 .4 .5 .6 130LET U1 = X5 131LET U2 = X7 132CHAR BLANK ALL; LINES SOLID ALL 133X1LABEL WIRE BRAID TYPE (X5) 134Y1LABEL WIRE DIAMETER (X7) 135X3LABEL 136CALL DEXCONT.DP 137X1LABEL 138Y1LABEL 139. 140. STEP 10--SORT Y AND CARRY ALONG X... FACTORS 141. 142ER 143LET Y2 = SORTC Y X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 X11 X12 X13 X14 X15 144SET WRITE DECIMALS 3 145WRITE X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 146WRITE X11 X12 X13 X14 X15 Y2 147WRITE OUT3A. X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 148WRITE OUT3B. X11 X12 X13 X14 X15 Y2 149. 150