Simulation for delay prediction.

We follow a object oriented approach for creating a general Queueing System and getting simulated wait time data. Further, we are also going to change the linear model to incorporate other ML techniques and hopefully create a better predictor.
Check out our code documentation here.(Currently in work).

Simulation Models
We follow a object oriented approach for creating a general Queueing System. Three main classes are declared in the components folder:

1. station : A single queueing system. You can define the number of servers (even dynamic i.e. changing with time). Also you can provide custom departure time distribution by providing the inverse distribution function ( pass a function { float -> float } to station class that generates the random service times). You'll have to use one the following constructor:

   station (long init_mxN, C_type C_para, event_type dept_para, float t=0, int init_n=0)
   
   station (long init_mxN, C_type C_para, float init_dept, float t=0, int init_n=0)
   
   station (long init_mxN, int init_C, float init_dept, float t=0, int init_n=0)
   
   station (long init_mxN, int init_C, event_type init_dept, float t=0, int init_n=0)
   

   Here C_type is Function type float -> int and event_type is Function type float -> float.
   Example of simple queueing system and more information on how to simulate can be found in examples folder.

2. tandem : You can pass in a vector of station objects to group multiple queueing system in series.

3. graph : Currently in work but the idea is to create a directed acyclic graph of station objects.

If you have already generated distribution for arrival or service times you can use,

std::vector<float> read_csv(std::string filename,int index); //index is the column to be read.

to read the data out of a csv file. and then the data can be used in our model in the following way:

std::vector<float> service_times = read_csv("lognormal.csv",1);
std::vector<float> interarrivaltimes = read_csv("lognormal.csv",2);

station MG1(1,1,
[service_times](float t) -> float
{
    float U = random;
    int index = (int)(U*service_times.size());
    try
    {
        return service_times[index];
    }
    catch(const std::exception& e)
    {
        // std::cerr << e.what() << '\n'; 
        return service_times[index-1];
    }
});

All test cases can be found here.

---

Project by-
Nalin Shani 2018ME10057 [email protected]
Achintya Eeshan 2018ME10094 [email protected]
B.Tech IIT Delhi Mechanical Engineering