part2 analysis
play

Part2: Analysis Prepared by: Paul Funkenbusch, Department of - PowerPoint PPT Presentation

Part2: Analysis Prepared by: Paul Funkenbusch, Department of Mechanical Engineering, University of Rochester Review ANOM ANOVA Error estimate Replication vs. Pooling ANOVA table Judging statistical significance DOE


  1. Part2: Analysis Prepared by: Paul Funkenbusch, Department of Mechanical Engineering, University of Rochester

  2.  Review  ANOM  ANOVA ◦ Error estimate  Replication vs. Pooling ◦ ANOVA table ◦ Judging statistical significance DOE mini-course, part 2, Paul Funkenbusch, 2015 2

  3. Example Terms (measure the volume of a balloon as a function of temperature and pressure)  Temperature,  Factors  variables whose influence you want to study. Pressure  Levels  specific values given  50  C, 100  C to a factor during experiments 1Pa, 2Pa (initially limit ourselves to 2- levels)  Set T = 50  C, P = 1Pa  Treatment condition  one running of the experiment and measure volume  Response  result measured  measured volume for a treatment condition DOE mini-course, part 2, Paul Funkenbusch, 2015 3

  4. Leve vel Use X1, X2, etc. to designate factors. Factor -1 +1 Use -1, +1 to designate levels 50 100 X1. Temperature (  C) X2. Pressure (Pa) 1 2 X1 at level -1 means T = 50  C y  response y = volume of balloon TC TC X1 X1 X2 X2 y Use a table to show factor levels and response (a) for each 1 -1 -1 y 1 treatment condition. 2 +1 -1 y 2 For example, during TC2, 3 -1 +1 y 3 set T = 100  C and P = 1Pa, 4 +1 +1 y4 measure the balloon volume = y 2 etc. DOE mini-course, part 2, Paul Funkenbusch, 2015 4

  5.  Test all combinations Leve vel Factor -1 +1  Responses  4 DOF 50 100 X1. Temp (  C) ◦ 4 measured (y) values X2. Pressure (Pa) 1 2  Effects  4 DOF ◦ 4 values calculated TC TC X1 X1 X2 X2 X1*X2 *X2 y ◦ m* = (y 1 +y 2 +y 3 +y 4 )/4 1 -1 -1 +1 y 1 ◦ D X1 = (y 3 +y 4 )/2 - (y 1 +y 2 )/2 ◦ D X2 = (y 2 +y 4 )/2 - (y 1 +y 3 )/2 2 -1 +1 -1 y 2 ◦ D 12 = (y 1 +y 4 )/2 - (y 2 +y 3 )/2 3 +1 -1 -1 y 3 4 +1 +1 +1 y 4  Model  4 DOF ◦ 4 constants in model ◦ y pred = a o +a 1 X 1 +a 2 X 2 +a 12 X 1 X 2 DOE mini-course, part 2, Paul Funkenbusch, 2015 5

  6.  Most basic level of analysis  Which effects are largest (which factors/interactions most important)?  Which levels produce the best (highest or lowest) responses?  Just based on the D values ◦ you’ve already done this     D    average response at level 1 - average response at level 1  m - m   1 1 DOE mini-course, part 2, Paul Funkenbusch, 2015 6

  7. Graphically Sign and Magnitude of D  Positive D ◦ +1 level should increase the response m* ◦ -1 level should decrease the response  Negative D  reversed A -1 A +1 B -1 B +1  Magnitude of D  Choose A -1 and B +1 to increase response ◦ Indicates relative importance  A is more important DOE mini-course, part 2, Paul Funkenbusch, 2015 7

  8.  Can treat interactions terms the same way  Larger D  more important interaction  If interactions are large  “best settings” to increase (or decrease) the response will depend on the combination of factor levels  Use model to test different combinations ◦ y pred = a o +a 1 X 1 +a 2 X 2 +a 12 X 1 X 2 DOE mini-course, part 2, Paul Funkenbusch, 2015 8

  9.  From part I. Leve vel Factor -1 +1  Removal rate of X1. applied load (kg) 2 3 osteotomy drills vs. applied load and X2. previous cuts 0 20 number of cuts. (new) TC TC X1 X1 X2 X2 X1*X2 *X2 Remov oval al rate  Which effects are (mm3/s) /s) most important? 1 -1 -1 +1 3  How can you 2 -1 +1 -1 2 increase the 3 +1 -1 -1 5 removal rate? 4 +1 +1 +1 2 DOE mini-course, part 2, Paul Funkenbusch, 2015 9

  10. Factor Leve vel  Effects -1 +1 ◦ D X1 = +1 X1. applied load (kg) 2 3 ◦ D x2 = -2 X2. previous cuts 0 20 ◦ D 12 = -1 (new)  Number of previous cuts is most important  new drill (level -1) will increase removal rate the most  Applied load and the interaction are comparable  higher load (level +1) will increase removal rate  but need to test combinations (since interaction is important too). DOE mini-course, part 2, Paul Funkenbusch, 2015 10

  11.  y pred = 3.0 + (0.5)X 1 - (1.0)X 2 - (0.5)X 1 X 2  Set X2 = -1 (Much larger than X1 and interaction)  What is best level for X1? ◦ For X1 = -1, X2 = -1  y pred = 3.0 ◦ For X1 = +1, X2 = -1  y pred = 5.0  Best settings X1 = +1 (3kg load), X2 = -1 (new drill)  Note; This is a synergistic interaction, ◦ Best level for interaction (-1) corresponds to best factor levels [X1X2 = (+1)(-1) = -1] ◦ Interaction enhances effects of “best” factor level choices  For an anti-synergistic interaction, ◦ Conflict between best factor settings and best interaction level ◦ Best overall settings then depend on relative strength of the interaction vs. factor DOE mini-course, part 2, Paul Funkenbusch, 2015 11

  12.  Second level of analysis  Which of the observed effects are statistically significant? ◦ Based on comparing observed effects against an estimate of error. ◦ Compares D 2 for factors and interactions with D 2 for error. ◦ Actually compare “mean square” or “MS”  proportional to D 2  How much does each factor/interaction contribute to the total variance in system? DOE mini-course, part 2, Paul Funkenbusch, 2015 12

  13. Pooling of higher-order Replication interactions Repeat (replicate) each of the Assume that higher-order   treatment conditions interactions are unimportant/zero Independent experimental runs Must choose these interactions   (not multiple measurements upfront (before examining results) from the same TC)  these form a “pool” for error Differences in the responses Effects measured for pooled   measured for identical TCs run interactions are used to estimate at different times provide error error “Pure error”  not dependent on “Error”  includes modeling error   modeling assumptions (i.e. assumptions about interactions) Best way to estimate error, but Requires less experimental effort,   greatly increases effort but error estimate is not as good DOE mini-course, part 2, Paul Funkenbusch, 2015 13

  14. TC TC X1 X1 X2 X2 y TC TC X1 X1 X2 X2 y 1a -1 -1 y1a 1 -1 -1 y1 2a -1 +1 y2a 2 -1 +1 y2 3a +1 -1 y3a 2X 3 +1 -1 y3 4a +1 +1 y4a 4 +1 +1 y4 1b -1 -1 y1b 2b -1 +1 y2b  Original design two factors at 2- 3b +1 -1 y3b levels  4 DOF  m*, D X1 , D x2 , D 12 4b +1 +1 y4b  Replicated design (2x)  + 4 DOF for error  8 DOF total  Contrast responses measured  4 DOF  m*, D X1 , D x2 , D 12 under nominally identical TC DOE mini-course, part 2, Paul Funkenbusch, 2015 14

  15.  Test more factors. Increase size and DOF. TC TC X1 X1 X2 X2 X3 X3 y 1 -1 -1 -1 y1  Example: three 2-level factors instead of two. 2 -1 -1 +1 y2  8 DOF total 3 -1 +1 -1 y3 1 DOF for m* 4 -1 +1 +1 y4 3 DOF for factors D X1 , D X2 , D X3 3 DOF for 2-factor interactions D 12 , D 13 , D 23 5 +1 -1 -1 y5 1 DOF for 3-factor interaction D 123 6 +1 -1 +1 y6  Pool all interactions 7 +1 +1 -1 y7  decide before examining results 8 +1 +1 +1 y8  assess m*, D X1 , D X2 , D X3 (4 DOF)  use D 12 , D 13 , D 23 , D 123 for error (4 DOF) Note: alternative choice (pool only the highest order , 3- factor,interaction) is possible, but only leaves 1 DOF for the error estimate  not desirable. For larger experiments this is not as much of a constraint (more higher-order interactions that can be pooled). DOE mini-course, part 2, Paul Funkenbusch, 2015 15

  16. Four 2-level factors Five 2-level factors  2 5 = 32TC = 32 DOF  2 4 = 16TC = 16 DOF ◦ m*  1 DOF ◦ m*  1 DOF ◦ factors  5 DOF ◦ factors  4 DOF ◦ 2-factor inter.  10 DOF ◦ 2-factor inter.  6 DOF ◦ 3-factor inter.  10 DOF ◦ 3-factor inter.  4 DOF ◦ 4-factor inter.  5 DOF ◦ 4-factor inter.  1 DOF ◦ 5-factor inter.  1 DOF  Pool 4 and 5 factor int.  Pool 3 and 4 factor int. ◦ Assess factors and 2-factor ◦ Assess factors and 2-factor and 3-factor interactions interactions ◦ 6 DOF for error estimate ◦ 5 DOF for error estimate DOE mini-course, part 2, Paul Funkenbusch, 2015 16

  17. “Pure error” Assess more (no modeling factors for same assumptions effort (or same needed) number of factors for less effort)  Best for:  Best for: ◦ large numbers of factors ◦ small numbers of factors ◦ systems with strong ◦ systems with large time/cost constraints on uncertainties experimental size DOE mini-course, part 2, Paul Funkenbusch, 2015 17

  18. Source ce SS SS DOF MS MS F p % SS A 100 1 100 20 0.011 56 B 50 1 50 10 0.034 28 AxB 10 1 10 2 0.230 6 error 20 4 5 -- -- 11 Total 180 7 -- -- -- 100  Typical way of presenting ANOVA results.  Explain each column so you can interpret these results.  Most software packages will output their analysis in some variant of this format.  Will also show how you can do the calculations. DOE mini-course, part 2, Paul Funkenbusch, 2015 18

Recommend


More recommend