OVERVIEW

This assessment involves writing a report that summarises a data science related investigation that you have conducted on data that you have collected yourself. The investigation must involve the main topics covered in the subject,  most  noticeably  data  pre-processing  (representation,  wrangling,  tidying)  and  exploratory  data visualisation using R/RStudio.

It  some  ways  this   is  a   merger  of  Assessments  on   Exploratory  Visualisation  and  and   Data  Pre-Processing, however  neither the dataset nor the pre-processing/exploratory steps to be carried out will be provided, you have to make independent choices and decisions.

You  will  need  to  find  your  own  data  using  good  practices.  Your   dataset cannot be  smaller  than   1000 observations  of  5  variables,  except  if  the  targeted  data  science  problem  to  be  addressed  relates  to spatial- temporal data, case in which less than 5 dimensions could be allowed.

Preferably, you should use a dataset relevant to your place of work. Do not use data from textbooks or from R

packages. Do not use  data  from  the   same   public  sources  that   have  been  used  in  the   subject  (e.g.  UCI repository). Do not use data from on-line prediction competitions such as kaggle. You can use public data, but the data should be appropriate for addressing a relevant data science problem.

You don’t need to solve this entire data science problem in your investigation, but you need to clearly indicate what the targeted problem would be about and how your project can contribute towards addressing it.

You have to write a report with details about the problem in question, the data, the methods, results, analyses and findings. You  might like to look online for  research  papers for examples of how to shape your  report. Obviously many of these papers will have undergone extensive work to collect their data, we don’t expect that for you.

We also don’t expect you to win a Nobel prize with this assessment. Ideally, you will be able to demonstrate that:

(a) you have grasped important concepts associated with this subject, most noticeably data  pre-processing and exploratory visualization; and (b) you can communicate your investigation in a formal written manner.

Regarding (a), we expect that your investigation will include at least six (60% or more) of the following topics:

1.   Data representation

2.   Unstructured to Structured data

3.   Data cleaning

4.   Type conversion

5.   Missing value imputation

6.   Gathering/Spreading

7.   Data subset selection and/or subsampling

8.   Group-based data summarisation

9.   Variable selection and/or transformation

10. Exploratory visualisation using ggplot2

Regarding (b), the main body of the report (containing title, abstract, introduction, data, methods, results and discussion, and  conclusions) cannot  exceed  5  (five)  A4  pages  in  12pt  Roman  style  font  using  single line spacing. A maximum of 5 (five) additional pages are allowed for bibliographic references and appendices with any supporting  material  that you  may want to  include  (e.g.  your  R  codes).  Therefore, your  report cannot exceed ten  (10)  pages  in  total.  Only the  main  body  and  references  will  be  formally  assessed for  grading, though the additional material can help clarify any issues that may arise during the marking process. Further details about the report structure are provided in the following section.

. Conclusions:  Final  remarks  about  the  key  achievements  of  the  investigations  and  what  makes  them “interesting” or “useful”, right now or for future work. Achievements or findings should be contrasted with the original objectives or hypotheses of the project. Make sure that you mention any limitations of your work here. Limit the conclusions to no more than two or three paragraphs.

. References. List the sources your investigation has drawn from. Note that all references should be referred to in the text.

. Appendices (optional): Add any supporting materials (possibly your detailed R codes) that might be useful to help assess your work.

FORMAT

The main body of the report must be presented in 12pt Roman style font on no more than 5 (five) A4 pages, using single line spacing. Either a single column or double column format may be used.

References and appendices can be listed on at most 5 (five) additional pages.

In total, the report cannot exceed 10 pages.

WARNING: only the main body and the references will be formally assessed and graded.

IMPORTANT NOTES

1.   The entire project must be accomplished using R/RStudio. Any calculations, visualisations, results, etc. produced using software other than R/RStudio (e.g. Excel, Tableau, etc.) is not accepted and therefore will not be assessed. Exploratory visualisation must use package ggplot2, rather than functions from base R or other R packages. The report itself can be written using a text editor of your choice (e.g. Microsoft Word or alike); R Markdown is also accepted, but it is not compulsory.

2.   If you opt to not submit your R codes appended to the report, the instructor and facilitators reserve the right to ask you to do so if more details or evidence are deemed required to properly assess your work.  Refusal  to  comply  with  this  requirement  may  incur  in  your  work  being  considered  as  not delivered.

A WORD ON PLAGIARISM AND SELF-PLAGIARISM:

Plagiarism is the act of using another’s words, works or ideas from any source as one’s own. Plagiarism has no

place in a University. Student work containing plagiarised material will be subject to formal university processes.

In case significant portions of your own previous work (e.g. a report for a related subject you did in this or any other university) is recycled in a way that it could be fully or partially graded twice (“double-dipping”), this is considered self-plagiarism and will not be tolerated.