Docker-compose

Our docker composed of 3 services :

• db : MongoDB used to store our data
• finance : Microservice that retrieve value from yahoo-finance API. We use these data as resources for each city
• api : Our server built using Express and json-api.

Build and run

First, you need to have docker and docker-compose installed on your machine.

After cloning the repository, simply run :

• docker-compose up

When all services are launched, the database will start stocking data providing by our microservice API.

You can access to the database using any adapted software (Robo 3T, Studio 3T) at this adress : localhost:27027.

Our API is available at localhost:1337/api. From this adress, you can also access to our GeneticAlgorithm API following this prototype :

• http://localhost:1337/algogen?start=[city]&end=[city]

You can use any city from this list for start and end values.

• Bordeaux
• Lille
• Lyon
• Marseille
• Nantes
• Paris
• Toulouse

Here are more inforamation about the algorithm itself.

Genetic Algorithm - link

Overview

This algorithm generates a path from depCity to arrCity by maximizing the resources harvested. Here is the prototype :

const algoGen = (depCity, arrCity, errorThreshold = 0.01, maxIteration = 100, nbChromosome = 15, mutationOdd = 0.1)

The depCity is not in the chromosome, but the arrCity is at the end of each. The size of each chromosome is random.

Description

Following are the main steps detailed :

Initialization

Creation of the first parents (which their size is random from 2 to 20).

Creation of children by Cross-Over

Creation of children by Cross-over at a random index from each of two random parents. We just have to maintain the maximum size a chromosome can have.

Mutation of children

Mutation Odd is set at 10%, but the last gene of each chromosome cannot mutate as it represents the arrCity.

Checking of chromosomes

Some additionnal rules to have better results :

1. Never the same city(~gene) twice in a row
2. No more than twice the same city in a chromosome

Computing scores of chromosomes

We mesure how many resources a truck harvest and the distance traveled. We suppose a truck have no cargo limit. Plus, a malus is applied if the distance traveled is over limitKM = 3000.

Updating moving average

We keep track of the score average of the last five iterations. The error is calculated as the difference between the current iteration and the mean of the last five averages, as percentage.

Selection

The new parents are selected as the 73% best chromosomes. (Because Why not ?).

Iteration

The exit condition required are both :

1. At least 5 iterations
2. An error < errorThreshold = 1%

Returned object

The object returned is composed of :

• the path (from depCity to arrCity included)
• the number of resources harvested (minus the malus if there is one)
• the distance traveled
• the number of iteration
• the final error

Credits

• Thierry Gonard : https://github.com/tuxnut
• Thomas Le Flohic : https://github.com/QuarkyUp
• Aurélien Vernizeau : https://github.com/Cerclique