Next Page Previous Page Home Tools & Aids Search Handbook
5. Process Improvement
5.6. Case Studies
5.6.3. Catapult Case Study

5.6.3.7.

Conclusion and Summary

DEX Contour Plot Important factors and best settings have been derived via the usual analysis approach. Important factors, in order of impotance, were X3 (number of bands), X4 (arm length), X1 (band height), X5 (stop angle), and weakly X2 (start angle). Some interactions (especially X3*X4) were also important. Best settings included:
  • For Y = 30: (xx) or augment the (yy) data
  • For Y = 60: augment the (xx) data
  • For Y = 90: (0 0 + 0 0)
Anomalies existed in the data. Plots revealed higher variability for X3 = +1 than for X3 = -1 implying a serious violation of the constant variation assumption for least squares fitting. Cross-validation revealed poor-quality fits for the X3 = +1 data, especially at the X3 = +1 pseudo-center point; hence interpolation was suspect. Corrective action is thus suggested:
  1. Transformations of Y (to correct non-constant variation).
  2. Subset Analyses (to correct poor-fitting regions).
The following section carries out such a corrected analysis.
Home Tools & Aids Search Handbook Previous Page Next Page