• Outline the steps to a data science analysis

• Understand the software requirements for the course

This content draws upon material developed by Jeff Leek & Roger Peng for their Advanced Data Science Course at the Johns Hopkins Bloomberg School of Public Health.

1. Define the question
   
2. Get the data
   
3. Clean the data
   
4. Explore the data
   
5. Fit statistical models
   
6. Communicate the results

1. Define the question
   
2. Get the data
   
3. Clean the data
   
4. Explore the data
   
5. Fit statistical models
   
6. Communicate the results

What makes a good Q?

• It should be specific
  
• It should be answerable with available data
• It should come with a definition of success
  
• It should not be designed to discriminate or cause harm

Data sources (in order of confidence):

• You collected your own
  
• Collaborators
  
• Public repositories

This is usually the most time intensive step

It’s also the most important & the hardest to teach

Do the data meet your expectations?

• Make judicious use of plots

• Also consider correlation matrics (eg, {corrplot})

• Beware of data mining!

Start with first principles

• Type of data (discrete, continuous)

• Distributions of data (normal, Poisson)

• Start simple and build from it!

Who’s your audience? What’s your medium?

• Lab group

• Social media, blog posts

• Presentation

• Publication

Consider these real scenarios

• Your collaborator asks, “Can you re-run the analysis with these new data?”

• “Your closest collaborator is you six months ago, but you don’t reply to emails.” (Karl Broman)

The goal is to explore the data

This is your “sandbox”

Soften the edges of the raw results

• Focus the analysis or “refine the arc”

• Improve the presentation

• Narrow your conclusions

A refined analysis and presentation

• Code in supplementary materials (GitHub, Zenodo, figshare)

• High quality figures

Here are some useful questions to ask yourself

• Could the question be answered with the available data?

• Did you understand the context for the question and the scientific application?

• Did you record the experimental design?

• Are the data “tidy”?

• Did you record the recipe for moving from raw to tidy data?

• Did you record all parameters, units, and functions applied to the data?

• Did you make univariate plots (histograms, density plots, boxplots)?

• Did you consider correlations between variables (scatterplots)?

• Did you check the units of all data points to make sure they are in the right range?

• Did you try to identify any errors or miscoding of variables?

• Did you identify the quantities of interest in your model?

• How did you select your model(s) (eg, information theory, cross validation)?

• Is this a correlative or causative study?

• Did you describe the data set, experimental design, and question you are answering?

• Did you specify the model you are fitting?

• Does each table/figure communicate an important piece of information?

• Did you create a script that reproduces all your analyses?

• Did you save the raw and processed versions of your data?

• Did you record all versions of the software you used to process the data?

• Did you try to have someone else run your analysis code?

We’ll learn about research compendia and style conventions for coding and naming