data-science-2
data-science-2
Anyway, what’s wanted for assignment 3? This is more about preparation for what your final project will be. You do not have to keep to the topic you discussed before. If you’ve found something more appropriate or appealing, use it instead.
A short paper, for the drop box, which says the following:
A. What is the subject of the envisaged project? A quick statement of what you are investigating and hope to demonstrate. For example, and I’m not saying that either of the following hypotheses are true, here are some fairly flippant things to look at: “clothing designers prefer brighter colors in years of economic growth” or “are there baseball teams that win more often on cold days than they win on hot days?”. (the first has references like http://www.pantone.com/pages/pantone.aspx?pg=21111&ca=90 and the second http://www.baseball-reference.com/). More serious examples would be “are we overfishing the world’s oceans?” or “the rise and fall of digital cameras.” (first starting with perhaps
http://www.ices.dk/marine-data/dataset-collections/Pages/Fish-catch-and-stock-assessment.aspx
and the second from https://petapixel.com/2014/12/15/chart-shows-badly-digital-camera-sales-getting-hammered-smartphones/).
Based on the earlier assignments a number of people are overly ambitious – it is not likely that you are going to be able to predict the stock market or the results of the NFL season in a few weeks of class work. Pick something that you can imagine describing in a few pages with a few charts, and will have some conclusion. It’s more important to have visualizations of your results than statistical measures. The late Richard Hamming (inventor of error-correcting codes) once said “the purpose of computing is insight, not numbers.”
You are welcome to use my examples for your project, or variations on them. But you should pick something you understand. If you can’t tell a herring from a halibut, you might want to avoid the fish project. I assume most people will do something they thought of themselves.
B. What are the data you are going to use? You should have found one or more datasets, with accessible numerical data. You may explain that you are going to have to do some editing of the data; I’m always going through files removing extraneous characters, reformatting, etc. If you aren’t familiar with editors or a language in which character editing is simple (i.e. not R, but Python or Perl), be careful to choose some data that doesn’t need much preprocessing. Please include in the paper some URL (or printed source) and the column headings (data schema) of what you want to use. In fact, include one sample row of data to be sure you can get through the basic download and extraction work.
C. What is an example of a conclusion about data that you like and might want to model your paper on? This can be a popular article or a scientific article, but something that uses the kind of data you are going to work with and says something about it. Find such an article and provide a brief summary and explanation of why it’s an example of using data for investigation or argument. Here are some examples which address important public policies and have gotten a lot of attention:
1. We used to put tetraethyl lead into all automotive gasoline. Did that cause increased crime in the 1970s-1980s? Kevin Drum popularized this hypothesis with impressive charts tracking crime. See
http://www.motherjones.com/environment/2016/02/lead-exposure-gasoline-crime-increase-children-health
for one of his articles and
http://pic.plover.com/Nevin/Nevin2007.pdf
for similar data in other countries. Note that this is politically controversial – if you start researching the subject you’ll find a lot of criticism of this hypothesis. My interest right now is the way he and his colleagues used data and graphics to make his point.
2. Did “austerity” (national budget cutting) hurt national economies, particularly in Europe, over the last ten years? Paul Krugman has written often about this, for example
https://krugman.blogs.nytimes.com/2012/04/24/austerity-and-growth-again-wonkish/
and
https://krugman.blogs.nytimes.com/2015/11/20/the-expansionary-austerity-zombie/.
Again, note his reliance on data and charts. This is even more politically controversial – there is an entire industry of economists attacking Paul Krugman.
3. Is energy production going to shift from fossil fuels to wind and solar? See for example,
http://reneweconomy.com.au/solar-output-uk-two-thirds-higher-coal-july-85288/
and a less convinced estimate (this topic is controversial as well):
http://www.eia.gov/todayinenergy/detail.php?id=10951
4. And here’s one that isn’t controversial: how can you classify music by style, and are children’s preferences in music changing over the last few decades?
http://www.icce.rug.nl/~soundscapes/VOLUME08/The_four_dimensions_of_popular_music.shtml
5. Finally, a self-serving topic: should you learn R to get a job?
http://blog.revolutionanalytics.com/popularity/
Anyway, your submission should include one (or more) examples of a paper or article in the topic area you have chosen that shows a conclusion based on some data. In this case I will say that if you use one of my examples find an article in addition to the ones I’ve given just above.
How long should this be? Well, perhaps 5 pages is a good target, But length is less important than having a topic, a data source, and a sample of somebody else who’s written about the area, with some comments on the data and the article.
