Pokémon GO
Wenjie ZHENG
Metagaming
Data Science
Today's talk
- Pokémon GO
- Metagaming
- Data science
Pokémon GO
- Augmented reality
- Mobile game
- Most popular in 2016
Catch a Pokémon
Gym Battle
Metagaming
- In-game info => gaming
- Out-of-game info => metagaming
In-game
- Combat power (CP)
- Hit point (HP)
- Moves
- Power
- Energy consumption
Out-of-game
- Attack
- Defense
- Stamina
- Moves
- Duration
- Energy generation
Information in Pokémon GO
Topics
- Pokémon tier list
- Strongest attacker/defender
- Gym battle team choice
1. Pokémon tier list
Three measures
attack
\text{attack}
attack×defense
\text{attack} \times \text{defense}
attack×defense×stamina
\text{attack} \times \text{defense} \times \text{stamina}
A
B
C
1. Pokémon tier list
1st
2nd
3rd
B
C
B
C
Before
After
1. Tier list (Count in DPS)
1st
2nd
3rd
B
C
B
C
After
Before
2. Strongest attacker/defender
Payoff matrix
2. Strongest attacker/defender
Before
After
Before
After
PCA on attackers (before)
2. Strongest attacker/defender
PCA on attackers (after)
2. Strongest attacker/defender
PCA on defenders
2. Strongest attacker/defender
3. Gym battle team choice
xijmins.t. i∑cij∑xij⎩⎨⎧∑iaijxij≥1∑jxij≤1xij≥0x’s row sparsity≤6∀j∈gym∀i∈inventory∀i,j
\begin{aligned}\min_{x_{ij}} & \sum_{i}c_{i}\sum_{j}x_{ij}\\
\text{s.t. } & \begin{cases}
\sum_{i}a_{ij}x_{ij}\ge1 & \forall j\in\text{gym}\\
\sum_{j}x_{ij}\le1 & \forall i\in\text{inventory}\\
x_{ij}\ge0 & \forall i,j\\
x\text{'s row sparsity}\le6
\end{cases}
\end{aligned}
Linear Programming
Difficult constraint
Gym battle team choice
xij,zimins.t. i∑cij∑xij⎩⎨⎧∑iaijxij≥1∑jxij≤zi∑izi≤6xij≥0zi∈{0,1}∀j∈gym∀i∈inventory∀i,j∀i
\begin{aligned}\min_{x_{ij},z_{i}} & \sum_{i}c_{i}\sum_{j}x_{ij}\\
\text{s.t. } & \begin{cases}
\sum_{i}a_{ij}x_{ij}\ge1 & \forall j\in\text{gym}\\
\sum_{j}x_{ij}\le z_{i} & \forall i\in\text{inventory}\\
\sum_{i}z_{i}\le6\\
x_{ij}\ge0 & \forall i,j\\
z_{i}\in\left\{ 0,1\right\} & \forall i
\end{cases}
\end{aligned}
Zero-One Linear Programming
Pokémon GO Wenjie ZHENG Metagaming Data Science