Please enable JavaScript.

Coggle requires JavaScript to display documents.

series 1 - Coggle Diagram

- - - - city_grid = np.full((GRID_SIZE, GRID_SIZE), None)
      - Agent is placed at the left top corner (0,0)
    - - Q_table = {}
      - Q table is used to see the best posible options
    - - It is use tu calculate the compatibility between building, so the agent knows which buildings are better together
    - - Represents the learning rate, determines the extent to which new information overrides old information
    - - Represents the discount factor, controls the importance of future rewards ( Makes the agent more patient or impatient).
    - - Represents exploration rate, controls the likelihood of random action (exploration) vs choosing best known action
    - - Build
        
        Residential
        
        Retail/Entertainment
        
        Green Areas
        
        Health Institutions
        
        Offices
        
        Educational institutions
        
        Supermarkets
        
        When building it places the building letter on the grid
      - It is able to choose which building to place on the grid (where it is at (0,0) )
        
        At this point the building is random because it does not have previous knowledge. (Q table is empty )
    - - All the values of the variables are the exact same for this stat with the exception of the city_grid, which contains the previous state building that was placed
      - Possible acctions
        
        Build
        
        Residential
        
        Retail/Entertainment
        
        Green Areas
        
        Health Institutions
        
        Offices
        
        Educational institutions
        
        Supermarkets
        
        When building it places the building letter on the grid
        
        It is able to choose which building to place on the grid (where it is at (0,x) )
        
        At this point the building is random because it does not have previous knowledge. (Q table is empty )
      - At the end of the state, the agent moves to the right to the next cell until there is no cell to the right
        
        when there are no cels left to the right it moves down and starts at column 1
      - x
        
        It represents the
        column of the grid
      - S(x,y)
        
        y
        
        represents the column in which the agent is placed
        
        X
        
        represents the row in which the agent is placed
        
        The agent repites the previous state and then goes down 1. It repeat the process until it ends on the lower right corner of the grid
        
        This state has the same posible actions as the states before
      - city_grid
        
        The grid contains the previous placed buildings of the other states
  - - - the more episodes that passes, the agent will make better decision on which buildings to place because it will have more accurate data on the Q table
    - - Represent the episode number of the state, at the end of an episode the x increases
      - There is a maximum value for x
        
        after x is maximum the agent goes to the next series
    - - it is calculated by the formula f(y) -f(x)
        
        f(y ) represents a function that calculates based on the series it is on ( it decreases its value the more that y increases )
        
        f(x) it is a number that will increment depending on the episode it is on
  - - - It is a function to calculate how well did the agent perform.
        
        It is based on the houses
        
        Makes a search of all the buildings that are connected to a particular house. They need to be connected by roads that are connected to the main road
        
        The main road needs to be placed at the center of the complete grid ( not only the chosen one ).
        
        If there is no road in this cell, the reward will always be 0
        
        each building that is close ( M cells ) to the house, will add a particular reward depending on the building and the amount of building of that type that are close
        
        particular_building_Reward Bonificacion Activado+ edificio2...+ edificio3
        
        Bonificacion is a dictionary that has the maximum amount of building of one type that are allowed to have connection with a particular house
        
        1 more item...
        
        Activado, means if there is connection to the house by the roads and main roads
        
        Particular_building_reward is a specific value for each building based on the compatibility matrix.
        
        M represents the Manhattan distance
        
        It is a variable that is placed by the user
- - - - The grid is n x n , n being the maximum cels