multiple linear and logistic regression

Question 1

What is multiple linear regression?

Answer 1

• models the relationship beyween one dependent variable and multiple independent variables using a linear equation

Question 2

What are the underlying principles of logistic regression?

Answer 2

• models the probability of binary outcomes by using a logistic function to transform a linear combination of inderpendent variables

Question 3

Advantages of linear regression

Answer 3

• explains relationships between variables
• can predict values of the dependent variable
• accounts for the influence of multiple predictors simultaneously

Question 4

Advantages of logistic regression

Answer 4

• predicts categorical outcomes
• outputs possibilities for classifications
• handles non-linear relationships via the logit transformation

Question 5

When is multiple linear regression used in data analysis?

Answer 5

• when predicting a continuous dependent variable based on multiple continuous or categorical independent variables

Question 6

When is logistic regression used?

Answer 6

• when predicting a binary outcome based on one or more independent variables (mutiple predictor values
• independent variables are casually related to dependent variable
• hypothesis testing (same as t test)
• importance of multiple variables predicting another (R^2)

Question 7

How can you recognise and interpret output from multiple linear regression?

Answer 7

• look for coefficients that indicate the strength and direction of relationships between predictors and the dependent variable, r squared for model fit and p values for statistical significance

Question 8

how to recognise and interpret output from logistic regression?

Answer 8

• analyse the odds ratios to understand the effect of predictors
• look at p values for significance
• use the classification accuracy or confusion matrix to evaluate model performance

Question 9

What is a key conclusion you can state from multiple linear regression output?

Answer 9

• if independent variables significantly predict the dependent variable and how much variance they explain

Question 10

what conclusion of logistic regression output can you state?

Answer 10

• which predictors significantly influence the probability of the binary outcome and the odds associated with these predictors

Question 11

What is a cofounding variable?

Answer 11

• unequally distributed but has an affect on outcome

Question 12

What are the advantages of multiple linear regression?

Answer 12

• able to adjust for confounding variables
• examine the effect of multiple independent predictors on an outcome
• improves the amount of the variability you can explain in the dependent variable
• interaction effects
• perform multiple hypothesis tests
• more accurate predictions of outcome variable