Category: Uncategorized

Under the Wald test, the coefficients of which variables are statistically significant at the 0.1 alpha level? Select ALL correct answers.

For logistic regression, if one or more model assumptions do not hold (i.e. no goodness of fit), the predicting variables have no predictive power.

Question 2: Bike Data – Full Model (2a) 2 pts – Using bike_data_train, fit a poisson regression model with bikes as the response variable and all other variables as predicting variables. Include an intercept. Call it model1. Display the summary table for the model.  (2b) 2 pts – Provide a meaningful interpretation of the estimated regression coefficient for precipitation for model1. (2c) 3 pts – Perform a test for the overall regression on model1. Is model1 significant overall using an alpha of 0.05? Why/Why not? 

Which statement below is NOT correct?

Question 6: Wine Data – Variable Selection (6a) 3 pts – Using wine_data_train, conduct a complete search to find the submodel with the smallest BIC. Fit this model. Include an intercept. Call it all_subsets_model. Display the summary table for the model.  Note: Remember to set family to binomial. (6a.1) 0.5 pts – Which variables are in your all_subsets_model?(6a.2) 1 pt – What is the BIC of all_subsets_model? (6b) 2.5 pts – Conduct backward stepwise regression on wine_data_train using AIC. Allow the minimum model to be a logistic model with quality as the response variable and only an intercept, and the full model to be model3. Call it stepwise_model. Display the summary table for the model. Note: Remember to set family to binomial. (6b.1) 0.5 pts – Which variables are in your stepwise_model? (6b.2) 0.5 pts – What is the AIC of stepwise_model?

The Lasso tends to yield sparse models – that is, models that involve only a small subset of the variables when a large number of predicting variables are considered in the model selection.

Using AIC and BIC criteria for model selection is suitable for linear regression, but not for generalized linear models such as Logistic regression and Poisson regression.

In Elastic Net, the  penalty encourages a grouping effect in the presence of highly correlated predictors.

Multiple Choice Questions 34-35 Below is an incomplete table of ANOVA for a linear regression model. Analysis of variance Source DF SS MS F-statistic Regression  1 920.45 920.45 ? Residual 8 18.91 ? Total 9 939.36 Using the above table, answer Questions 34 and 35.  

Which of the options below is NOT a common sign of multicollinearity in linear regression?