MSc in Statistics University of Sheeld MAS6002 Statistical Laboratory 2012/13 Assignment 3 Assignment 3 is in two parts not related to each other. Your submission in MOLE should combine reports on the two parts into a single pdf document. Your work in both parts should be based in R. Deadline: 12 noon Tuesday 18th December 2012. Part 1 The i-th deletion residual e??i is dened as e??i = yi ?? xTi^ ??i where xTiis the i-th row of the design matrix X and ??i is a column vector of least square parameter estimates calculated without the i-th observation.
MSc in Statistics University of Sheeld MAS6002 Statistical Laboratory 2012/13 Assignment 3 Assignment 3 is in two parts not related to each other. Your submission in MOLE should combine reports on the two parts into a single pdf document. Your work in both parts should be based in R. Deadline: 12 noon Tuesday 18th December 2012. Part 1 The i-th deletion residual e??i is dened as e??i = yi ?? xTi^ ??i where xTiis the i-th row of the design matrix X and ??i is a column vector of least square parameter estimates calculated without the i-th observation. Write some annotated R code to calculate the deletion residuals when the linear model yi = 0 + 1xi + 2x2i+ i is tted to the data in the le quadratic.txt on MOLE2. By drawing suitable plots comment on the distribution of these deletion residuals. There is no length constraint on part 1. Part 2 The data in the table below give the average temperature in degrees Fahrenheit for the rst week in January of the years indicated for a number of cities in the USA along with approximate latitudes. Treat latitude as a continuous variable and year as a categorical variable in any analysis you do. City Latitude Year 1966 1967 1968 Fargo 1 10 9 -5 Marquette 1 25 22 9 Burlington 1 30 12 16 Lincoln 2 34 18 18 Peoria 2 35 23 16 Columbus 2 42 29 22 Atlantic City 2 39 31 27 Oklahama City 3 51 31 30 Memphis 3 52 37 32 Asheville 3 42 34 34 Hatteras 3 54 44 47 Austin 4 60 45 43 New Orleans 4 60 53 46 Tallahassee 4 60 53 49 (i) By tting some appropriate linear models discuss the evidence for an interaction between latitude and year in this data set. 1
(ii) The command a.lm

Attachments: