This repository analyzes the tab data from HWS USUDC 2020 to see the effect of tapered scoring on the break. This README summarizes the process and results; for more details on the analysis, read analysis.md.
I collected the draw from every round of USUDC by downloading each draw's webpage as a PDF, then extracting the draw in a tidy, tabular form from the PDFs. (Some portion of the PDF tables were inaccessible within R, so I supplemented this with the OCR software Tabula.) All this raw data is contained in the raw_data directory. In the script create_results.R, I combined these tables to create the clean dataset final_data/results.csv.
In addition to the draw, I downloaded the tournament break and extracted it to final_data/breaks.csv.
My analysis has two results:
- Each round had no statistically significant effect on a team's probability of breaking.
- Judge quality in round 1 does not distort the break.
Each of these is expanded on in analysis.md. To inspect the code and analysis itself, see analysis.Rmd.