Learning Goal: I’m working on a r exercise and need an explanation and answer to

Learning Goal: I’m working on a r exercise and need an explanation and answer to help me learn.Suppose I have a long-format data-frame df, and clean it with na.omit, then perform correlation in it via cor, (i.e. cor(na.omit(df))). What is the equivalent code in R? Cor (df, use=”p”)
Cor(df, use=”complete.obs”)
Cor (df, na.omit=TRUE)
Cor (df, na.rm=TRUE)
What is an acceptable way to calculate a linear regression of y as a function of x in R? glm(x~y, family=”gaussain”)
glm(x~y, family=”linear”)
glm(x~y, family=”simple”)
None of the Above
What is true about the Box-Cox regression optimization routine? Box-Cox will correct for non-linearity, but does not guarantee homoscedascity
Box-Cox will correct for heteroscedascity, but does not guarantee linearity
Box-Cox will guarantee a linear, homoscedastic scatter when done
Box-Cox does not guarantee linearity, nor homoscedasticity when done
What is the impact of two collinear variables in a multiple regression? They will mutually redundant and may corrupt the explanatory power/variance
They will fail the Breushpagan test for homoscedascity
They will fail the shapiro-Wilk Test for normality
All three of the above are true
When implementing a power transformation (PT), you could conceivably transform both you x- and y data. But, what is the caution against this (i.e. why should you transform just one variable and not both)?PTs change the units, too, so interpretation is made more complex
PTs are hard to implement: multiple PTs could prolong the computation by hours
Every time you PT, you surrender some Type-1 error, so do so sparingly
The Best practice is: PT the first variable; Log-transform the second
Requirements: 1-2 sentence explanation

Leave a Reply

Your email address will not be published. Required fields are marked *