Based on the below tables, which model should be chosen among five? Please expla

Question

Based on the below tables, which model should be chosen among five? Please explain your answers in terms of AIC, McFadden R2 , Area under the ROC curve, and Model Complexity (i.e., Number of Predictor Variables).

Training

Model 1

Model 1.1

Model 1.2

Model 2

Model 3

AIC

9242

11299

11226

2772.8

9235.8

McFadden R-squared

0.334

0.184

0.190

0.610

0.334

Validation

Model 1

Model 1.1

Model 1.2

Model 2

Model 3

Area under ROC

0.9010

0.7903

0.7955

0.9019

0.9012

Number of Predictor Variables

20

12

14

19

15

Training

Model 1

Model 1.1

Model 1.2

Model 2

Model 3

AIC

9242

11299

11226

2772.8

9235.8

McFadden R-squared

0.334

0.184

0.190

0.610

0.334

Explanation / Answer

The best model should have the following characterisitcs :

1) It should have lowest AIC
2) It should have highest McFadden R-square
3) It should have highest Area under ROC
4) It should depend on less number of predictors.

From all the 5 models provided in the table it is clear that model 2 satisfies criteria 1 to 3. From view point of complexity of model it is ranked 4th. But due to the other factors mentioned above, it is legit to choose model 2 as the best model.

Related Questions

Navigate

• Browse (All)
• Browse B
• Subjects

---

---

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.