badminton-court-scheduler

Use npm run dev to test locally.

Scheduling Algorithm

The court scheduling algorithm aims to optimize groups of players on the available courts based on the following heuristics:

1. TIME: Players who have waited longer should have higher priority
2. DIVERSITY: Players who have played together less should be paired together with higher priority
3. SKILL: Players should be matched with approximate balance between the two sides

The algorithm assigns courts with the following logic:

1. Order all players by their TIME priority. This is the 'queue'.
2. Perform a greedy algorithm to select the players to add to the next court (players are moved to the end of the queue once they are selected):
   1. Assign the first player in the queue to team 1
   2. Pick a player to join team 2: check the next 'x' players in the queue and pick the one with the highest combined TIME and DIVERSITY score with the selected team 1 player
   3. Pick team 1's partner: check the next 'x' players in the queue and pick the one with the highest combined TIME and average DIVERSITY score with the other 2 selected players
   4. Pick team 2's partner: check the next 'x' players in the queue and pick the one with the highest combined TIME and average DIVERSITY score with the other 3 players, such that the total SKILL scores of both teams match the target difference.
3. Repeat step 2. as many times as needed to fill the "upcoming games" display.

Time Heuristic

The app should track the time each active player finished their last game. Players who have waited longer will be ranked higher in the queue.

Diversity Heuristic

The app should track a list for each player, that maps to all other active players (basically, a matrix of players). Each entry should store the time since these players last played together, and/or the number of times these players have played together. Players who have not played together in a long time, or who have played together few total times will be paired together with higher priority.

To make this more robust, we could also separate this value into "played on same team" and "played against each other", which could be weighted separately (so people who have only played together would be weighted higher to play against each other).

Skill Heuristic

Players will be assigned a skill rating in their player data (assigned by the admin or by themselves depending on setup choice). The algorithm uses this to balance teams on each court, so that the numbers are close to matching on each side. For example, we could store player skill in tiers, like (1, 2, 3) and the algorithm would try to keep both sides at the same total value (so a 1 and 3 against 2 and 2 for example).

To add variety into this, since it's often fun when there's occasionally imbalance in games (being the "underdog"), we could have this balancing follow more of a curve, where it normally tries to balance the teams exactly, but also makes some games that are imbalanced by 1 or 2.

Furthermore, we could potentially add much more depth into the skill rankings, such as splitting it into sub-skills, making a more precise scale (out of 10 or 100), or other such metrics. Custom player preferences could also potentially be applied here (pairing couples together? etc.)