Week 05: Puting it all together#
Time to put everything you learned in this class into action. You will prepare an Exploratory Data Analysis Project. In an exploratory data analysis project, you are just looking at (hence exploring) the data and learning about the data and the possible relationships between variables.
This is not meant to be a formal statistical analysis. You cannot make any claims about groups being statistically different or causally related. This is just an excuse to practice what you learned by describing a dataset of your choice. You are allowed and encouraged to hypothesize why you observe certain relationships between variables in your chosen dataset. Just try to resist making causal claims from what is most likely an observational dataset.
Instructions#
Using your data set of choice, pose a brief research question that explores the relationship between 2 - 3 variables. Use Markdown headers to make the following sections.
Introduction: A short introduction/description of the data.
Specifically mention the 2-3 variables you are going to explore.
What is your research question? What are you interested in finding out more about?
Univariate Exploration: Describe each of the variables under consideration.
This means calculate some summary statistics (N(%) or mean(sd)) and make a graphic
Bivariate Exploration: Comparison between two variables of interest.
Calculate grouped summary statistics as appropriate. This is often the most often forgotten part
You can go further and explore more than two variables at a time using paneling, but be sure to explain what you learn from each graph.
Conclusion: What did you find? If you had a prior hypothesis, does the data seem to support it? Remember this is NOT a statistical analysis.
All descriptions (univariate and bivariate) must be done using graphics, summary statistics, and sentences.
This is a very vague set of instructions for a reason. You are to choose variables and explore anything that you find interesting. Create tables, graphics, grouped summary statistics (mean of the continuous variable across levels of the categorical variable). Whatever you need to do to understand the relationship between these two measures.
Use the grading rubric at the bottom of this page for guidance as to what you should present, in what order, and the level of detail you need to present.
Data#
You have a choice here. If you are currently working on some data that you would like to explore, talk with your instructor to get your data set approved. As long as it has more than a few variables in it, and at least 30 observations it should be fine.
If you do not have your own data, you can choose from one of the following data sets, all of which can be downloaded from the Data page of Dr. D’s teaching course website. Here are some viable choices:
Email Spam: Characteristics of emails used to predict if the email is spam or not.
HIV: Data on adolescent children living with HIV positive parents.
Depression: Level of depression (cesd), health care, and demographic characteristics. High School and Beyond: Educational, vocational, and personal development of elementary and high school students.
Police Shootings: Characteristics of individuals killed by police in 2015.
Any other datasets, and/or any that we’ve already used in class require instructor approval.
How to submit#
Each student is to create a Google Colab notebook and set the Share permissions to General access Anyone with the link.
A Google spreadsheet will be created and shared with everyone. Each student will have their own row. You are to paste the URL to your Colab notebook in your row.
Deadline#
To allow for adequate time for peer grading the submission deadline is a strict cutoff. Late assignments won’t be accepted. It is better to submit something, than nothing.
Guidelines#
Run all, early and often. To ensure all of the code you develop in your Colab notebook works as expected, it is highly recommended to click Runtime > Run all, essentially after you finish code for each next code cell.
Proof read your report prior to submission. Please read and re-read your report, editing for clarity and removing any duplicate content (sentences and/or code).
Hide output of any code that is not necessary for your reader. Put a semi-colon
;on the last line of code within a code cell to reduce excessive printing. Nobody needs to read through raw data afterall.This is to be independent work. Papers that are too similar will receive no credit. Look at the grading rubric to help you decide the level of detail required.
Grading rubric#
The criteria below is what you will be graded on.
Data Description. Provide a description of the data set and the variables of interest.
Does not meet expectations. There is no description or the description is a copy of the data metadata.
Meets expectations. There is a minor description of the data but not enough to understand what is being measured or compared.
Exceeds expectations. The data description is clear and concise and it is clear to me what data is being analyzed and where it was obtained.
Univariate Description. Fully describe the distribution of each variable by itself.
Does not meet expectations. There are no numerical or graphical summaries provided.
Meets expectations. Only numeric or only graphical summaries were created, but no textual description.
Exceeds expectations. The variable was compared using appropriate numerical and/or graphical methods and short textual explanations immediately followed.
Bivariate Comparison. Describe the relationship between the two chosen variables.
Does not meet expectations. No comparisons were made, or the comparisons were inappropriate/incorrect.
Meets expectations. The variables were compared using appropriate graphical methods and grouped summary statistics were created, but nothing was discussed.
Exceeds expectations. The variables were compared using appropriate numerical and/or graphical methods and short textual explanations immediately followed.
Organization / Grammar. How well does the report read? How well organized was the report? Was it checked for grammar and spelling mistakes?
Does not meet expectations. Only Python code and output are present. There is no discussion of results. Lots of extraneous code that is not relevant to the discussion is present. Markdown was used inappropriately.
Meets expectations. An attempt was made to discuss the results, but the explanations are not in a report format or there are some large grammar and/or spelling problems. Some Python code that is not relevant to the analysis question at hand is being displayed. Markdown headers were used to create sections.
Exceeds expectations. The report was spell checked and written well, with full English sentences. The report flowed well and followed the required order of discussion topics with Markdown headers used successfully.