GLM Mastery

What is the main concept behind Generalized Linear Models (GLM)?

• Fitting a linear regression model to the data

• Using non-linear transformations on the predictors

• Modeling the relationship between the mean of the response variable and predictors through a link function and a probability distribution.

• Ignoring the distribution of the response variable

Can you name a commonly used link function in GLM?

• exponential

• sigmoid

• linear

• logit

How would you interpret the coefficients in a GLM?

• The coefficients are only relevant for continuous predictor variables.

• The coefficients represent the probability of the response variable being zero.

• The coefficients indicate the strength and direction of the relationship between the predictor variables and the response variable in the GLM.

• The coefficients are always positive in a GLM.

What are some ways to evaluate the model fit of a GLM?

• Examining residuals, checking goodness-of-fit statistics, conducting hypothesis tests on model coefficients, and using diagnostic plots.

• Relying solely on goodness-of-fit statistics

• Using only one type of diagnostic plot

• Ignoring residuals completely

In what scenarios would you apply GLM to analyze data?

• GLM is applied when the relationship between variables is linear

• GLM would be applied to analyze data when the response variable follows a specific distribution and the relationship with predictors is not linear.

• GLM is suitable for analyzing image data

• GLM is used for text analysis

Explain the role of the link function in GLM.

• The link function in GLM connects the linear predictor to the distribution of the response variable.

• The link function in GLM determines the sample size of the data.

• The link function in GLM is only applicable to continuous response variables.

• The link function in GLM is used to calculate the residuals.

How do you identify the appropriate link function for a given dataset?

• Use the same link function for all datasets

• Choose randomly

• Ask a friend for their opinion

• Consider the nature of the response variable: identity function for continuous unbounded, logit function for binary/Bernoulli, log function for count/Poisson.

What does the deviance statistic measure in GLM?

• It measures the difference between the model with only the intercept and the full model with predictors.

• It evaluates the significance of the intercept term.

• It measures the accuracy of the model predictions.

• It calculates the mean squared error of the model.

Describe the difference between linear regression and GLM.

• Linear regression only works with categorical variables, while GLM only works with continuous variables.

• Linear regression always requires interaction terms, while GLM never does.

• Linear regression assumes normal error distribution and a linear relationship between variables, while GLM allows for non-normal error distributions and non-linear relationships.

• Linear regression is used for classification problems, while GLM is used for regression problems.

When would you use a Poisson distribution in GLM?

• When modeling count data

• When analyzing survival data

• When dealing with categorical variables

• When modeling continuous data