Logistic Regression Group

Logistic Regression Report

Data, Method, and Model

Number of cases used in the analysis 23
Number of incomplete (omitted) cases 0
Link Function logit
Model cbind(damaged, undamaged) ~ temperature

Response Information

Variable Value Count
damaged Damaged O-ring 11
Undamaged O-ring 127
trial_var Total 138

Parameter Estimates

Model: cbind(damaged, undamaged) ~ temperature
Event = Damaged O-ring

Term Coefficient Standard Error Z-Value P-Value
(Intercept) 11.663 3.296 3.538 0.0004028
temperature -0.21623 0.05318 -4.066 4.777e-05

Odds Ratios

Model: cbind(damaged, undamaged) ~ temperature
Event = Damaged O-ring

Term Change Odds Ratio 2.5% Lower CL 97.5% Upper CL
temperature 1 0.8055 0.7258 0.894

Measures of Model Fit

Model: cbind(damaged, undamaged) ~ temperature
Event = Damaged O-ring

Measure Value df
Deviance* 16.9123 21
Null Deviance** 38.8977 22
Log Likelihood -14.8374 2
MSE (Brier Score) 0.00913606
R-Squared 56.52%
Adj R-Squared 54.45%
AIC 33.6748
BIC 35.9458

* Deviance for the specified model.
** Deviance for the intercept-only model.

Goodness-of-Fit Tests

Test DF Chi-Square P-Value
Deviance 21 16.9123 0.7164
Pearson 21 28.0674 0.1383
Hosmer-Lemeshow* 6 7.38052 0.2871

* Number of groups used for the Hosmer Lemeshow test = 8

Confusions Matrix

Event = Damaged O-ring; Classification Cutoff = 0.5

Actual Predicted Damaged O-ring Predicted Undamaged O-ring Total
Damaged O-ring 5 (3.62%) 6 (4.35%) 11 (7.97%)
Undamaged O-ring 1 (0.725%) 126 (91.3%) 127 (92%)
Total 6 (4.35%) 132 (95.7%) 138 (100%)

ROC Plot and Cost Function Plot

ROC and Cost Plots, AUC = 0.824
 FP Cost = 1, FN Cost = 1, Optimal Cutoff = 0.345, Total Cost = 7for the ROC curve , ,
    The x-axis is False Positive rate and it ranges from 0 to 1, ,
    , the y-axis is the TRUE positive rate, and it ranges from 0 to 1.
    For the cost function the x-axis is the classification cutoff and
    the y-axis is the cost that ranges from 7 , , to , ,127