Which one of the following statements about GLMs and GAMs is FALSE?

a.An iterative algorithm is used to fit both GLMs and GAMs. That is, it requires repeated computations that converge to the fitted model

b.If a data frame is fed into glm() using the data argument then if there are missing values in any variable in the formula then, by default, that row is deleted

c.The glm() function and gam() function(s) models the mean response

d.For GAMs the coefficients of smooth terms are highly interpretable (like LMs) and may be extracted by coef()

e.For GLM and GAM software the fitted() and predict()generic functions return values  and η^i respectively by default

f.Using a log link with binary data is possible but has the danger of fitted values being greater than 1

Smoothing requires the variable being smoothed to have certain properties. Alternatively, having certain properties may not be required but are certainly good if they hold. Which one of the following properties definitely need not hold?

a.Must be finite

b.Preferably has many unique values, e.g., more than 10

c.It's best that that variable is stored in a data frame

d.Must be numeric

e.Must be positive valued

Given vectors x (sorted and unique) and y, and a single value dfval, suppose one fits

fit <- smooth.spline(x, y, df = dfval, all = TRUE)

resss <- sum((y - fitted(fit))^2)

Suppose one steadily increases the value of dfval. Which one of the following statements is TRUE?

a.resss will decrease in value only if x is sorted into descending order

b.resss will remain unchanged

c.resss will decrease only if x is sorted in to ascending order

d.It is impossible to say how resss will change

e.resss will decrease in value

f.resss will increase in value

Answer rating: 100% (QA)

If a data frame is fed into glm using the data argumen... View the full answer