代写syntax、代做java/c++、代写data BGSall、CS/python编程代做

日期：2018-12-10 10:25

Intructions and Recommendations:

Please start your document with the following statement: “On my honor, I have not had any form of

communication about this exam with any other individual (including other students, teaching assistants,

instructors, etc.)”.

Your answers should be presented in the first 2 pages of your submission paper. You are strongly

encouraged to have succinct answers. Lengthy answers are neither required nor desired (and they are

prone to contain incorrect statements that would be penalized).

There is no unique correct solution and is unlikely that two students who haven’t work together have

the exact intermediate steps and final solution. As a consequence, your grade is not based on the final

solution, it is based on

– a proper and adequate use of the techniques/methods learned in class.

– appropriate arguments made when making decisions, whether or not those decisions required the

use of statistical methods.

– Appropriate supporting material (syntax/charts/output and any lengthy explanations after the

first two pages) that support your work and conclusions. There is no limit of space for supporting

material.

You are advice to avoid methods not covered in class. If however, you believe this is necessary (because

for example, that would lead you to select a very different model that is clearly more appropriate), then

you need to attach somewhere on your document a small explanation of the external method used and

provide meaningful references (i.e., papers, manuscripts, or textbooks are valid references; Wikipedia or

stackoverflow are not).

Submit your solutions to Canvas as a single PDF file by Monday, December 10th. Do not wait until

the very last minute since late submissions may not be graded.

1

Questions

Use the data BGSall These are the research questions we want to answer:

a. Could information about weight and height from individual’s earlier years help determine their body

mass index at age 18, BMI18?

b. Assume that the data is representative of a given society. Assume that we are interested in avoid

problem with obesity for the adult population but funding could be provided to improve nutrition in

only one of the following groups in that society: i) infants up to 2 years of age, ii) children older than 2

but younger than 9, or (iii) children older than 9 but younger than 18. Is there evidence, based on

these data only, that it would be better to provide funding at earlier stages (i), intermediate stages (ii),

or later stages (iii)?

1. Use help(BGSall) to get familiar with the data at hand.

Why wouldn’t be appropriate to use height at 18, HT18, and weight at 18, WT18, to study this

problem Explain.

Would it be appropriate to separate boy and girls in two different data sets, or deal with them

jointly including a factor to account for this difference? Justify your answer and if the former,

answer the following questions twice (one for each group).

Is there any non-statistical argument based on the research questions or your own knowledge that

you could use to select a subset of regressors from your entire data set? If so, use it to simplify the

set of predictors before your start your data analysis. If not, use all the predictors in the data set.

2. Use graphical methods and statistical techniques learned in class to study your data and come up with

a final model. Show your proficiency checking for transformations, interactions, polynomials, etc., even

if it’s clear you may not use some of them later on.

3. Based on your work in question 2, come up with an “almost final” full model, and use any valid

arguments (statistical or non-statistical) to compare it with one (or few) reduced model(s), and conclude

with a candidate model.

4. Using your final candidate model, check to see if there are any clear violations to the assumptions of

the model. Also check if there is evidence of collinearity. Make appropriate corrections, if any, and

briefly interpret key components of the summary output of your model.

5. Based on your final model in question 4, answer research questions (a) and (b). Question (a) is straight

forward. For question (b) choose only one of the following options:

Your results help answer this question. Explain how.

Your results are not enough to answer this question but with a few additional considerations you

could use them. Give an example what those considerations could be and provide an answer based

on them and your model.

Your results are not useful at all to answer this question. Explain why.

6. Based on your final model in question 4, perform an influence analysis, in particular

using the Cook’s distance, determine which observations are the most influential (select up to

four),

determine which observations, if any, could be considered outliers,

compare the coefficient estimates obtained with and without the selected influential observations

and briefly comment on your results.

is there any conclusions that you would change in questions 5 due to the results obtained here

Explain.