Data Science with Python

How to think like a data scientist

Creative Commons License

Knitr

Assignment

Learning Objectives

Following this assignment students should be able to:

  • use knitr and rmarkdown to make documents with reproducible analysis
  • refresh your ability to use obtain, manipulate, and graph tabular and spatial data

Reading

Lecture Notes

  1. knitr

Exercises

  1. Reproducible Breeding Bird Survey Analysis (100 pts)

    You are interested in understanding how the biodiversity of birds varies in response to environmental variables and decide to conduct your analysis in a reproducible manner using knitr and rmarkdown. Specifically you want to know how species richness (the number of species seen at a site) varies in response to the mean annual temperature and the mean annaual precipitation.

    1. Start a new Rmd document with a title and author and set the output format to html_document.
    2. Add a markdown chunk that describes the question you are going to ask.
    3. Add a code chunk that loads the required packages. Hide the output from loading packages using message = FALSE.
    4. Add a header related to downloading and importing the data.
    5. Add a text section briefly describing the two datasets you are going to use.
    6. Add a code chunk to download the Breeding Bird Survey data using the rdataretriever package. Instructions for installing this package and the associated Python package are available on the Data Retriever website. It will take a long time to download and convert this data into a set of useable CSV files (~30 minutes), so add a conditional statement that checks to see if the necessary files have already been created and only install they data if they have not. Don’t show the output for this chunk.
    7. Add a code chunk to load the species, counts, and routes tables into R and display the top few rows of each table.

    8. Make a map of the locations of all of the Breeding Bird Survey routes, including an outline the North America landmass. Add a header above this map describing what it shows. You can get a world map useing usmap = map_data("world"), which you can then plot using geom_polygon. To only show this data in the region of the Breeding Bird Survey routes add the following to you ggplot command:

      scale_x_continuous(limits = c(min(routes$longitude), max(routes$longitude))) +
scale_y_continuous(limits = c(min(routes$latitude), max(routes$latitude)))
    9. Use the getData function from the raster package to obtain the bioclim data (getData('worldclim', var = 'bio', res = 10)) and extract the values for each route. Convert resulting matrix into a data frame and select just the mean annual temperature (bio1) and the mean annual precipitation (bio1). Use cbind to combine these two predictor columns with the routes table.
    10. Determine the species richness at each route in 2015. To get unique routes you will need to group by by the statenum and route columns. Join this data with the predictor data you obtained in (7).Display the new table data.
    11. Make two graphs, one each showing the relationship between bio1 and richness and bio12 and richness. Include the raw data points and a smooth line through them. (optional) Try doing this with a function if you want an extra challenge.
    12. Write a brief conclusions section providing your interpretation of the results.
    13. Return to the data section of your document and add citations for both datasets. You will need to create a .bib file to hold your bibtex citations. You can obtain bibtex for the citations by searching Google Scholar for “Breeding Bird Survey” and “Worldclim”, clicking on the " icon, and selecting bibtex. You should also add a References header at the bottom of your document since the references will appear at the end.
    [click here for output]

Assignment submission & checklist