3) (20 points) In the PUMS NY data that we’ve been using in class, those with TRANWORK == 70 are working from home

3) (20 points) In the PUMS NY data that we’ve been using in class, those with TRANWORK == 70 are working from home. Compare that group of people with those commuting on the subway (there’s a dummy Commute_subway or use TRANWORK == 33). What are the educational attainments in each group? Given that someone works from home, what is the likelihood that the person has at least a 4-year degree? Given that someone is female, what is the likelihood that she works at home? Given that someone is male, what is the likelihood that he works at home? Create a confidence interval for the difference and provide a p-value.

4) (40 points) With the same data, create a different regression where TRANTIME is the dependent variable (you can drop the ones with zero values). What is the effect (if any) of educational qualification on commuting time? What about other demographic effects such as age and race/ethnicity? Explain confidence intervals and p-values for each important coefficient (where your explanation determines important). Create a joint hypothesis test to find a p-value for whether all of the education dummies are all equal to zero. Calculate predicted values for some people and assess if these seem plausible. Perhaps a graph of data and predicted values.

5) (20 points) Create a k-nn model to try to predict how a person commutes to work. (Worry a bit about the zero values of TRANWORK since that typically means they’re not working.) How useful is this model at predicting? What are some of the important variables in this prediction?