STAT375 Linear Models Assignment 3, 2018

Linear Models Assignment代写 Instructions•You should submit the assignment using the Turnitin tool on iLearn.•When finding the “best” model, do not give

Due date: Tuesday 29 May, 1pm

Instructions Linear Models Assignment代写

• You should submit the assignment using the Turnitin tool on
• When finding the “best” model, do not give the outputs of all of the models that you consider. You should give a table summarising your model selection process, and write downyour final  Give the output of the final model in an appendix.

Question 1 Linear Models Assignment代写

Afifi and Azen (1979) report data which were collected on 113 patients admitted to the emer- gency department at Los Angeles County Hospital with the diagnosis of shock,  a potentially  fatal condition.  There are six different types of shock reported in the data set, each of which   has a different cause.   Measurements of some physiological variables were made at the time     of admission. In addition, whether the patient died in hospital or survived is recorded. The following variables are in data file shock assn3 2018.sav:

Variable Description

death 0=survived, 1=died

shock shock type (2=non-shock, 3=hypovolemic shock, 4=cardiogenic shock, 5=bacterial shock, 6=neurogenic shock, 7=other)

age age in years

sex 1=male, 2=female

SBP Systolic Blood Pressure (mm Hg) MAP Mean Arterial Pressure (mm Hg)

We are interested in investigating how patients’ mortality depends on the covariates observed.

a.Investigate the data graphically and numerically, and comment on the relationships that you observe.Linear Models Assignment代写

b.Suppose we wish to explain the odds of death, in terms of the covariates. Using a model selection criterion, find the “best”

c.Foryour best model, give an interpretation of the effect of each of the covariates in your model, on the odds of death. Do these agree with your investigations in (a)?

d.Using your finalmodel: Linear Models Assignment代写

i.Compute the fitted odds of death of a patient with the followingprofile:

ii.Hence compute the predicted probability that a patient with the above profile does not survive.

e.Givethe classification table, using a cut-off of the prior probability of  What is the sensitivity and specificity?

f.Construct the ROC curve and give the area under the curve. What does the area under the curve indicate in thiscase?

Reference Linear Models Assignment代写

Afifi, A.A. and Azen, S.P. (1979). Statistical Analysis (A Computer Oriented Approach), 2nd edition, New York, Academic Press.

Question 2

The data set absent assn3 2018.xlsx contains data on the number of days absent in a school year, of a sample of n = 316 high school students from two schools. Variables are:

Variable Description

daysabs number of days absent

maths standardised score on mathematics test lang standardised score on language test gender 0 = female, 1 = male

school 1, 2 Linear Models Assignment代写

Develop a statistical model for the number of days absent. You should investigate the data graphically before fitting a model. Use a model selection criterion to select the best model. For your final model:

a.Write down the model

b.  i.Interpret the modelparameters.

ii.Which students tend to be absent from school?