Sector grouping

Display :

Description :
This module creates groups (commercial sectors for example) which have the same weight.
The algorithm begins with input poles, then it will extend the groups to have an equal repartition of input data in them.

Four methods are proposed :

• geographical extension
  This method generates groups by extending a group to its neighbours until the associated value is close to the theoretical value (total sum of reachable entities divided by the group number).
  The groups are correctly distributed on the map but the result can be far from the ideal solution.
  We advise you to use this method on map which elements have homogeneous values and poles are well distributed. The poles should be the entities with highest values.
  
  
• geographical extension with balancing
  The balancing enables to compensate the errors produced by the geographical extension. It will restore a balance in assigning the group entities which have a total value upper than the theoretical value (too big groups) to the ones that have a value lower than the theoretical value (too small groups). This balancing is done on the entities of the biggest groups that are neighbours with small groups.
  
  
• intelligent search among all solutions
  This method enables the best solution to be found by calculating all the possible solutions, but with an algorithm more selective than in the exhaustive search among all solutions.
  It will not run on maps with entities which look like "neck", that is they have a compulsory crossing between two entities groups.
  
  
• exhaustive search among all solutions
  We advise you to use this method on map which have less than 20 entities because the algorithm calculates all the possible solutions and keep the best one. The interest is to find the solution nearest of the theoritical solution on small maps.
  It will not run on maps with entities which look like "neck".
  
  

Each method uses a complex algorithm that provides neither a single solution nor a guaranteed result. Therefore, we advise you to save the generated data (thanks to the button in the data table).

If the Island connection box has been checked, then islands will be considered as neighbours of their nearest entity. You should check this box if your map contains islands (for example in the France by departments with the Corse).

If you uncheck the No limitation on treatment time box, you can specify the maximum time allowed for the chosen method to find a solution. This time is in seconds.
If the method can't find a solution because of a too small allowed time, then a message will appear and the supplied solution will be the best one found in this time.

The groups will have about a similar weight. An error rate enables you to know the reliability of the solution. The more the error rate is close to zero, the more the groups have a similar weigth.

You can choose the quality associated to groups : it can be the identifier, the name, or the expression identifier-name of each pole.

In the Result text zone, you can find the following informations:

• the group number
• the sum total of values for all map entities
• the objective for each group
• an error rate
• for each group:
  • the number of concerned elements
  • the associated value

Script :

2      module  untyped_list ""
3        mod_type  integer "103"
3        mod_subtype  integer "529"
3        mod_name  string "Sectorisation"
3        mod_dads  integer_list ""
4          ? integer  "8"
4          ? integer  "5"
3        island_process  boolean  "T"
3        weight_method  integer "802"
3        time  integer "30"
3        unlimited_time  boolean "F"
3        time_stop boolean "F"
3        output_id  boolean  "F"
3        output_name  boolean  "T"
3        output_id_name  boolean  "F"
3        group_tree untyped_list ""
4          group_nb integer "3"
4          sum_val double "543.965400"
4          group_value double "181.321800"
4          error_rate double "1.694211"
4          group untyped_list ""
5            grp_id string "AIN"
5            grp_elem_nb integer "30"
5            grp_value double "184.381000"
4          group untyped_list ""
5            grp_id string "AISNE"
5            grp_elem_nb integer "36"
5            grp_value double "176.713840"
4          group untyped_list ""
5            grp_id string "LOT-ET-GARONNE"
5            grp_elem_nb integer "30"
5            grp_value double "182.870560"
4          bad_poles string_list ""

Values for weight_method :
geographical extension                     801
geographical extension with balancing      802
intelligent search among all solutions     803
exhaustive search among all solutions      804