Descriptive Analysis Project

Complete by: Thursday 22 Oct. at class time

Assignment Description

You’ve already learned a lot about how to work with data, and this assignment is designed to get you started bringing all these skills together on a dataset you care about. The Documentation Assignment was stage one of the semester-long final project; this is stage two.

At this stage of the project, you’ll (1) choose a dataset to work on, (2) craft some research questions around that data, (3) conduct some secondary research on your data, (4) consider the ethics of your project, and (5) employ some exploratory data analysis to answer your research questions. The assignment will take the form of a Jupyter Notebook report that asks and answers some research questions about a dataset of your choosing. Let’s begin by learning a little more about the semester-long project.

The Final Project

The purpose of the final project is to combine the core skills you have gained in the class to produce a polished report on a question of your choosing, ideally something that you’re passionate about or is relevant to your life or interests.

This final project will incorporate all of the steps involved in the Data Analysis Cycle. You will be in charge of stating interesting questions that don’t duplicate previous projects or workshops, conducting exploratory analyses using skills from the entire semester, building models that you will interpret, then communicating your key findings in a polished, professional narrative. You can refer to the full prompt for the Final Project for more.

Project Structure

This stage of the project, the Descriptive Analysis, should be a Jupyter Notebook report with the following sections:

• a written introduction, which explains the dataset and the research questions you seek to answer
• a section on the ethics of your data and analysis, using the ADSA’s four lenses
• a short annotated bibliography of 2–3 sources relative to your type of data or mode of analysis
• at least 4 polished visualizations with clear descriptions
• at least 5 univariate or bivariate analyses of key attributes in your data
• a short conclusion that summarizes your findings

The report should be organized with findings split into different sections. To turn in the project, upload the HTML version of your Jupyter Notebook to Sakai. More details on specific elements of the project are below.

By Thursday 24 Sept. at 5pm, you should send me an email with the dataset you plan on using and some first ideas for research questions.

Project Elements

Selecting Data

You can choose whatever real-world dataset you like, but I recommend that you choose one of the three datasets that you found at the beginning of the semester, since you’ve been thinking about those for the last several weeks. If you need help finding datasets, refer back to the prompt for the Documentation Assignment.

Consider that your dataset will need to have certain features in order to conduct certain analyses. Most of our analysis methods work best when you have a fair bit of data, so you probably don’t want a dataset that’s too small. You should choose a dataset while keeping in mind the techniques and methods that we’re learning in this class: this will make it easier for you to choose models and approaches later on. You’ll use the same data for this project and the last stage of the final project at the end of the semester.

The way to know you’re choosing the right dataset is to have good research questions that will guide you. Mostly, choose something that you are curious about and will allow you to demonstrate the listed concepts for the project. Have some fun with it!

Asking Questions

You will want several research questions (a minimum of 3) that help you to address different parts of your analysis, as well as a rationale for how these questions relate to each other. Consider this textbook chapter for more on creating good research questions.

We can ask questions that are closed (yes or no answer; are the two things different or not?) or open (require more thought and explanation; how much does something change? How is it related to something else?). Likewise, our data analyses can serve one or more purposes as we move through the data analysis cycle: descriptive (describes different measures of the data), exploratory (looking for patterns or unknown relationships in the data); inferential (using a sample to tell us something about a larger population); predictive (use relationships in the current/past data to predict the future); or causal (what happens to one variable when one or more other variables change?).

Ethical Considerations

Consider some of the ethical challenges that your data presents, and write a paragraph discussing these. Remember: all data projects involve careful consideration of ethics. Address the ethical issues in your project in terms of the ADSA’s four lenses. You’ll be able to include an updated version of this paragraph in your Final Project. It is fine to adapt some of what you wrote in your documentation assignment, but in this section of your project, the ethical considerations should apply to your analysis, not only to your data collection.

Secondary Research

To help you craft your research questions, you’ll find two or three secondary sources that are relevant to your data or analysis. These could be existing studies on similar data, articles about particular methods, or summaries of data analysis in your chosen domain. You should include a formal citation (MLA is fine) for each source, briefly describe each one in a sentence or two, and then explain how they relate to your research project.

You can use the many resources in the Data Science LibGuide to help you find sources.

Your Descriptive Analysis

You should organize your analysis section based on the research question or questions you’ve created. You might have different sections based on different findings or different parts of the questions you’re answering. The exact structure is up to you, but this part of the analysis must include the following:

• the data wrangling steps you need to take to prepare your data
• at least 4 polished visualizations with clear descriptions
• at least 5 univariate or bivariate analyses of key attributes in your data

No more than two of your visualizations should be the same visualization type. Write text that describes the data and what the visuals tell you about your research questions. The statistical analysis may be summary statistics, confidence intervals, distributions, correlations, hypothesis tests, or other observations. These may be closely related to the visualizations you create in this section. Like in all of your reports, everything should be clearly described and interpreted.

Good luck, and remember that your project will change and grow as you work on it! Get started early, and let me know your questions as they come up.