Team: Natalie Dean and Xavier Beech
Inspiration
We were inspired by our own interests in and involvement with research at the University of Washington, along with a desire to learn more about the national landscape surrounding research institutions.
What it does
Takes multiple files of biennial survey results from research institutions and visualizes trends with geopandas and matplotlib. We showed trends in number of institutions per state, most common research subjects, and funding sources and amounts per state.
How we built it
We developed in parallel using GitHub, in Atom and VSCode.
Challenges we ran into
Data cleaning took a lot more time and effort than initially expected, especially with multiple datasets involved. We also wrote very generalized code between us that could potentially be used for many similarly-formatted datasets, so the streamlining of the code took significant amounts of time.
Accomplishments that we're proud of
We're especially proud of the structure of this project. We built a robust data cleaning and standardizing file that would allow for any number of files, if available to be put into this project should a future user desire. The code that we built in our data_analysis, and overall, was very tight and generalized and we're proud of the way it was written. This code could allow with few modifications for many different types of analysis, and would let the user look at years and patterns even outside of those we explored in our report.
What we learned
We learned that projects often take much more time than originally estimated, and to budget our time and expectations accordingly. We also learned about the importance of contextualizing data instead of looking at it in its raw form, as while a dataset on its own can tell the viewer a lot, getting the full picture from a small snapshot can be very difficult.
What's next for Assessment of US Science & Engineering Research Institutions
Further research is required to fully contextualize this data. Population data would probably be most useful, but sub-state resolution geodata would also be useful. These would be useful for things like per-capita calculations and analyzing accessibility to research institutions for citizens of the state.
Log in or sign up for Devpost to join the conversation.