To provide more value for the candidates in the competition, my task this week was to generate graphs which will be sent to them by email, that depict how well they did in relation to others. I started with using a TSV parser to pass the spreadsheet I generated earlier as input to produce the graphs.

Then I created histograms of the scores by venue, by instrument and by category using R. Next I added an additional point or line to show the candidate’s mark against this distribution, to show how well he/she did relatively.

This presented a problem, because the actual marks shouldn’t be shown. Hiding the axes helped with masking the marks, but still a candidate would be able to glean an estimate of his/her mark from looking at the graph. Thus I switched to using percentiles to generate the graph.

To make the graphs more visually appealing and intuitive to interpret, I switched to using density plots, which shows a curve matching the ‘shape’ of the histogram.

The graphs are generated as PNGs, which I then use libHaru to embed them into a PDF. Then this PDF is attached to an email and the email is sent to the candidate.


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.