LogisticRegressionLabSolutions

# LogisticRegressionLabSolutions - Recitation #11 and...

This preview shows pages 1–3. Sign up to view the full content.

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Recitation #11 and Homework #9: Logistic Regression — solutions OR&IE 3120, Spring 2009 Problem 1 As can be seen in the boxplots below, the variables RiserDiam and SprueDiamBot are most related to Feasible . RiserDiam is higher when Feasible is 1 and lower when Feasible is 0. SprueDiamBot is lower when Feasible is 1 and higher when Feasible is 0. An initial recommendation is to keep SprueDiamBot as small as possible and RiserDiam as large as possible in order to maximize the probability of a feasible casting. There is some subjectivity when answering this question. One might select SprueDiamTop , RiserHeight , or Riser2Pos as a third variable that is related to Feasible . Each of these is higher on average when Feasible is 1 compared to when Feasible is 0, so the initial recommendation would be to keep these mold-variables as large as possible. 1 0.2 0.5 0.8 GateDiam 1 4.0 5.0 6.0 CupHeight 1 9.0 10.5 SprueHeight 1 5.56.57.5 RiserDiam 1 0.3 0.6 SprueDiamBot 1 0.4 0.8 1.2 SprueDiamTop 1 3.5 4.5 5.5 RiserHeight 1 2.5 3.5 Riser1Pos 1 4.0 5.0 6.0 Riser2Pos 1 Problem 2 As seen in the summary below, at level 0.05, RiserDiam , SprueDiamBot , SprueDiamTop , and RiserHeight are statistically significant. Also, Riser2Pos is just on the borderline with a p-value just above 0.05, specifically 0.0503. Increasing SprueDiamBot decreases the proba- bility that a casting is feasible. This result agrees with what we saw in the boxplots. Increasing RiserDiam , SprueDiamTop , RiserHeight , or Riser2Pos increases the probability that a casting is feasible. These results also agree with the results in the boxplots. AIC is 96. Call: glm(formula = Feasible ~ . - BatchTime, family = binomial, data = cast) Deviance Residuals: Min 1Q Median 3Q Max-2.3206-0.5218-0.0818 0.5986 1.9538 Coefficients: Estimate Std. Error z value Pr(>|z|) (Intercept)-40.15784 11.96095-3.357 0.000787 *** GateDiam-1.54130 2.34685-0.657 0.511341 CupHeight 0.85936 0.84322 1.019 0.308135 SprueHeight 0.02714 0.54616 0.050 0.960372 RiserDiam 3.01801 0.80194 3.763 0.000168 *** SprueDiamBot -12.66655 3.35007-3.781 0.000156 *** SprueDiamTop 4.43038 1.91395 2.315 0.020625 * RiserHeight 2.57208 0.89799 2.864 0.004180 ** Riser1Pos-0.68042 0.85734-0.794 0.427405 Riser2Pos 1.37366 0.70168 1.958 0.050267 ....
View Full Document

## This note was uploaded on 11/01/2011 for the course ORIE 3120 taught by Professor Jackson during the Spring '09 term at Cornell.

### Page1 / 7

LogisticRegressionLabSolutions - Recitation #11 and...

This preview shows document pages 1 - 3. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online