4.1 Procedure
4.1.1 Data Collection
We have to develop a timetable using ACO variant MAX-MIN Ant System. For developing the timetable, firstly we need the timetable data of a university. So, we try to collect the data of DEI university. In DEI, there are two academic sessions: odd-semester and even semester in a year. Odd semester runs in the months of July to December and runs the classes of odd no semester, e.g.; first, third, fifth and seventh semesters. And even semester runs in the months of January to June and runs the classes of even no semester, e.g.; second, fourth, sixth and eighth semesters. Collecting the timetable data of whole DEI requires more time. So, because of less time, we collected the data …show more content…
4.2 Technology Used
4.2.1 Matlab
Matlab is defined as a high level programming language which supports all kinds of operation like preprocessing of the data. Then based on the preprocessed information analysis of data takes place and the analyzed data can be represented either statistically or graphically easily. This language can be used in wide variety of applications and is easy to understand.
In our work, we have to work with many matrixes to construct the timetable. And matrices required many operation. So, we need a technology which supports matrixes and performs operations on it. Matlab is a good tool which supports matrixes very well.
We have done all of the programming in matlab and make the timetable in matlab.
4.2.2 Microsoft Excel
Microsoft Excel is easy to use and we can easily store and manipulate the data in it. To make the timetable, we need to store the data of a university courses and the rooms. So, to store the data of timetable, we use the Microsoft Excel.
4.3 Snapshots
4.3.1 Details of Courses
Figure 4.1: Course Details
4.3.2 Details of …show more content…
The MAX-MIN Ant System (MMAS) algorithm has successfully generated the timetable of different sizes. The MMAS algorithm does not take more time for generating the timetable of small and medium size but it takes some time for generating the timetable of large size.
5.1 Parameter Tuning
We have determined the parameter values of the algorithm for different sizes of the UCTP problem experimentally by taking some test cases. By these parameter values in different UCTP problem instances, MMAS algorithm converges more rapidly with minimum no of soft violations and generates the timetable.
In the experiment, the most influence factor is the heuristic information (β) and the other influence factors are pheromone evaporation rate (ρ), pheromone weight (α) and no of ants (A) and no of iterations (I).
Figure 5.1: Parameters used in the different size of UCTP
Parameter Small Medium Large
Pheromone weight (α) 1 1