Spencer131
Veteran
why does event similarity matter?
event similarity should be considered in the debate of adding and removing events. if events are super similar, then they can be repetitive and redundant. to make competitions more fun and fair, we should have a diverse set of events
how can we measure event similarity?
in this post, I measure event similarity by looking at the rankings. if people have similar rankings across events, then I consider those events to be similar to each other
I did this by using the python programming language and the pandas library. I loaded everybody's ranking for each event into a dataframe, and only took people who have competed in every event. at the time of writing, this left 555 people.
I also divided everybody's rank by the maximum rank to normalize the data. normalization is needed because, for example, rank 100 is 3x3 is much better than rank 100 in 5bld
then, I loaded the ranks into a dataframe and used the corr function to compute the correlation between events
10 most similar events
here are the 10 most similar events with the correlation coefficient
10 least similar events
here are the 10 least similar events with their correlation coefficient
event similarity should be considered in the debate of adding and removing events. if events are super similar, then they can be repetitive and redundant. to make competitions more fun and fair, we should have a diverse set of events
how can we measure event similarity?
in this post, I measure event similarity by looking at the rankings. if people have similar rankings across events, then I consider those events to be similar to each other
I did this by using the python programming language and the pandas library. I loaded everybody's ranking for each event into a dataframe, and only took people who have competed in every event. at the time of writing, this left 555 people.
I also divided everybody's rank by the maximum rank to normalize the data. normalization is needed because, for example, rank 100 is 3x3 is much better than rank 100 in 5bld
then, I loaded the ranks into a dataframe and used the corr function to compute the correlation between events
10 most similar events
here are the 10 most similar events with the correlation coefficient
- 777, 666, 0.956
- 444, 555, 0.930
- 555, 666, 0.926
- 555, 777, 0.882
- 444bf, 555bf, 0.881
- 333, 444, 0.872
- 444, 666, 0.834
- minx, 555, 0.827
- 333oh, 444, 0.822
- 333, 333oh, 0.814
10 least similar events
here are the 10 least similar events with their correlation coefficient
- 333mbf, 666, 0.123
- 333mbf, 555, 0.122
- 333oh, 555bf, 0.116
- 555bf, clock, 0.111
- 555bf, 777, 0.109
- 333mbf, minx, 0.103
- 333, 555bf, 0.099
- 555bf, 222, 0.093
- skewb, 555bf, 0.089
- 777, 333mbf, 0.088
- pyram, 555bf, 0.053
