Data
The data I chose to use are the usage statistics for Smogon University's battle simulator, Pokémon Showdown!, which can be found here. This information is used to determine Smogon's tier listing, a representation of the viability of each individual Pokémon, and is openly provided for viewing by the public. Permissions to use the data were granted by the data collector in multiple posts, an example of which can be found here. Additionally, the Pokémon Showdown! battle simulator falls under the MIT License, while the code used to obtain stats by the collector falls under the GNU General Public License 3.0.
The original dataset contains the usage information for 53 months, beginning in November of 2014 and currently ending in April of 2019 at the time of writing. Each month contains upwards of 900 individual files examining various tiers and stats. In my initial processing I decided to focus on Smogon's most popular tier, OU, as well as narrowing down the month to August of 2016. This month was chosen for a multitude of reasons. By August 2016, Pokémon Omega Ruby and Alpha Sapphire had been released for almost two years, meaning that enough time had passed to where a very stable and consistent metagame could form. August is also both the month in which the Pokémon World Championships are held, as well as the final month before Smogon's fall tier shift, meaning that it both has data from an even higher number of players and carries more weight than some other months.
I further processed this data using a Java program that read in the human readable files, and parsed them to extract the relevant information and place them in csv files. A copy of this program can be found in the repository for this project. The resulting 6 files have from between 50 to 500 rows, as the data was filtered to show the top 50 Pokémon within the OU tier, which roughly aligns with Smogon's own tier cutoff of 3.41% usage.
