LOSS FUNCTION     The Loss Function offers a way to quantify the improvement from the optimum design determined from an experimental design study.  Definitions:     L = K (Y – Yo)^2   ….  for a single sample.    L = K (MSD)    ……..  for the whole population.   where      L    = Loss in dollars.                K    = Proportionality constant.                Yo  = Target value of the quality characteristic.                 Y    = Measured value of the quality characteristic.   THE COST SAVINGS WHEN THE MEAN VALUE IS HELD AT A TARGET VALUE CAN BE CALCULATEDWHEN THE FOLLOWING INFORMATION IS AVAILABLE :   – TARGET VALUE OF QUALITY CHARACTERISTIC.  – TOLERANCE OF QUALITY CHARACTERISTIC.   – COST OF REJECTION AT PRODUCTION (PER UNIT).  – UNITS OF PRODUCTION PER MONTH (TOTAL).        – S/N RATIO OF THE OLD DESIGN.  – S/N RATIO OF THE IMPROVED DESIGN.    : Since the S/N ratio is a direct product of ANOVA,     it is conveniently used for the calculation of loss.    However, the loss function requires MSD and must be calculated from the S/N ratio.

 

Special experiment designs.

 COMBINATION DESIGN  This is a method to fit two “2-level” factors in a “3-level” column. Suppose you have factors A and B at two levels and factors C, D, and E at three   levels.  An L-9 has four “3-level” columns. Factors C, D & E can occupy three columns leaving one column for A and B. A and B form A1B1 A2B1 A1B2 & A2B2. Select any three of these four and assign them to the three levels of the respective columns reserved for A and B.     DESIGNS TO INCLUDE NOISE FACTORS (OUTER ARRAY)This version (version 4.7) of the program simultaneously handles inner and outer arrays. The noise conditions for repetitions can be studied by    describing the outer array following the completion of the experiment design (inner array). Whether an outer array is present or not, up to 35 repetitions of results (columns) can be entered and an analysis performed using this software.