Before we do proper data science, it's important to run some exploratory data analysis. In particular, we care about the following variables.
Alias | Papers | StartYear | Stirling Diversity Index | h-index
Alias is a letter+number, to protect our subjects, papers is numbers of publications, StartYear is their earliest publication, the Stirling Diversity Index is from the last post, and the h-index is a common measure of academic success, where H is the highest number of H papers that have been cited at least H times. So for example, the h-index of a person's who publications have been cited (7,6,3,1,1,0) is 3, from the papers that have been cited 7,6, and 3 times.
I created an anonymized data file, 3-scientometrics101_EDA.csv, to use going forward.
And as always, the first step is to just take a look at the data with .head(), .describe(), and a quick pairplot. One thing to note is a few of the people we sampled aren't in the dataset at all. They didn't have any papers indexed by Web of Science.
Alias | Papers | StartYear | Stirling Diversity Index | h-index
Alias is a letter+number, to protect our subjects, papers is numbers of publications, StartYear is their earliest publication, the Stirling Diversity Index is from the last post, and the h-index is a common measure of academic success, where H is the highest number of H papers that have been cited at least H times. So for example, the h-index of a person's who publications have been cited (7,6,3,1,1,0) is 3, from the papers that have been cited 7,6, and 3 times.
I created an anonymized data file, 3-scientometrics101_EDA.csv, to use going forward.
And as always, the first step is to just take a look at the data with .head(), .describe(), and a quick pairplot. One thing to note is a few of the people we sampled aren't in the dataset at all. They didn't have any papers indexed by Web of Science.
A few things that immediately pop out are the three outliers, scientists who have authored over 200 papers. The straight line in the start year data around 2005 is something that I've seen before, and is probably an artifact. I think that in 2005 Web of Science began indexing many more journals. It looks like there's a linear relationship between papers and the h-index, which is not surprising. And I can't see any clear way to distinguish interdisciplinary and non-interdisciplinary scholars.
Plotting the histograms of SDI scores together lets us see that there might not be as much difference between the two groups in our sample as we thought. It may be time to pause and re-evaluate.
Or, since this is a proof-of-concept, press on regardless.
Or, since this is a proof-of-concept, press on regardless.
RSS Feed
