ST300解题步骤

ST300 Assessed Coursework 2019
Guideline to Layout and Miscellaneous
This document provides a brief guideline on what should be included in the
written report and the general layout, followed by miscellaneous remarks and
things you need to be careful at.
1 Layout
There should be a cover page (not counted towards the six pages of
A4 limit). Include the coursename and code (ST300), date and time
handed in (to the drop-box in the 6th floor waiting area) and your
candidate number (NOT your student number starting with 200x...).
Do not write your name on it.
For the main content, divide your report into named sections. For
example, see
http://www.maths.manchester.a... (if clicking
it directly does not work, copy and paste the link directly to a web
browser, it should work.) You may first want to describe briefly the
variables you have and the purpose of the whole analysis. Then give a
brief summary of your conclusion.
Try to make it clear your conclusion first after the description of variables.
The general idea is to find a model which is simple yet of “enough”
explanatory power to let you draw a conclusion on how the performance
is related to salary. You may have done a lot in order to find the final
model that you are satisfied with, but the general reader of a report
only need to know briefly what you have done in order to reach the
final model. Yet you need to give adequate details and justifications
of removing variables, removing data point(s), transforming particular
variable(s), etc. You may leave some of the graphical output related
to the process of reaching the final model to an appendix, and just
describe what you have done and the justifications in words in the
main content.
February 11, 2019 ST300 1
RGLM
If you have done something unusual and you think this achieves something
important, you may want to include the R code as well on top
of a description and justification of what exactly you have done. E.g.,
a fancy transformation (I recommend you do not go to far into finding
a transformation. Remember more complicated transformation may
hinder interpretability of the model).
Outliers in a linear model may not be an outlier anymore in a linear
model with transformed variables. If you are searching for a transformation,
remember to include all data points first. Then look for outliers
in the transformed model. Of course remember to provide justifications
(graphical outputs, values of suitable residuals, etc) for removing data
point(s) for your final model.
Give adequate descriptions and justifications for your final model, including
R outputs or related graphical outputs. Convince the readers
why this is a good model (e.g. regression diagnostics, ANOVA table,
hypotheses testing comparing with potentially interesting smaller or
larger model(s), etc). If you have transformed your response variable,
remember to interpret the model in the original scale.
Remember you only include something relevant in the report. Don’t
treat the appendix as a garbage can and put everything there. You only
have 6 pages, and appendix is for those things which are also relevant,
but may not be good to include in the main flow of the presentation.
Or, you can put some potentially interesting but unsuccessful analyses
there, which is a part of the process of reaching the final model.
2 Remarks and Miscellaneous
Remember to drop your report in the coursework drop box on the
6th floor of Columbia House, and hand in the signed plagiarism form
separately to the department office, both by 12:00pm on the 20th of
March 2019. Penalties will apply for late submission, at 0.5 mark each
day (10 marks total).
If you request an extension to the deadline, you need to fill in an extension
form. The link to the form is on Moodle below the project links.
Retrospective extension requests shall not be granted.
February 11, 2019 ST300 2
RGLM
Keep a copy of the report in case things go wrong and you need to
submit again.
Avoid font changes unless the text merits it. Avoid fancy fonts or
colours or special effects which will distract from the text.
Use plain English without jargon. Remember to run spell check, and
check the whole report again.
Use the first person plural (We) rather than singular (I), or write in
the passive voice (The model was examined...)
Eliminate any unnecessary materials from the report.
WX：codehelp