Greetings everybody
Does anyone have any idea of how I can evaluate whether my genetic algorithm is working well? what are the criteria?
Best Regards
Dear Amin Bakhtiari,
In the case of single objective GA, please refer to the CEC2005/2013 which contain problem definitions and evaluation criteria for the single objective real-parameter optimization problems. Otherwise, refer to the CEC2009 multiobjective optimization test instances.
http://dces.essex.ac.uk/staff/zhang/MOEAcompetition/cec09testproblem0904.pdf
http://al-roomi.org/multimedia/CEC_Database/CEC2013/RealParameterOptimization/CEC2013_RealParameterOptimization_TechnicalReport.pdf
https://www.lri.fr/~hansen/Tech-Report-May-30-05.pdf
Dear Amin Bakhtiari,
In the case of single objective GA, please refer to the CEC2005/2013 which contain problem definitions and evaluation criteria for the single objective real-parameter optimization problems. Otherwise, refer to the CEC2009 multiobjective optimization test instances.
http://dces.essex.ac.uk/staff/zhang/MOEAcompetition/cec09testproblem0904.pdf
http://al-roomi.org/multimedia/CEC_Database/CEC2013/RealParameterOptimization/CEC2013_RealParameterOptimization_TechnicalReport.pdf
https://www.lri.fr/~hansen/Tech-Report-May-30-05.pdf
Thanks alot dear Hojjat. You have given me exactly what I was looking for.
Dear Lov,
Im about to do a kind of multi objective optimization. To shed light into that, I have to day that I want to find the appropriate value of volume fraction of some nanofluids which are used to enhance heat transfer ability of parabolic trough collectors.
Dear Amin Bakhtiari ,
I have no any idea about your project. I suggest you to first define objective function. You already mentioned that you are working in multi objective optimization, then you have minimum two objectives function. After that you may write your own code or use matlab inbulid function.
These are the step to find the volume fraction.
1. Initially, you considered 50 chromosome which contain the value of your parameter.
2. compute the fitness values of each chromosome.
3. Sorting the chromosome based on their fitness value.
4. check the stopping criteria.
5. cross over and mutation performed.
6. go to step 2.
Dear Lov,
Thanks alot for your precise answer. Actually, I have done all of these and my algorithm is ready and working! My question is that How I can evaluate my code correctness? Well, as you can see, Mr. Hojjat Rakhshani has suggested 3 papers which contain some test suits by which it is possible to validate your algorithm. But this question again brings up that is this a reasonable way to validate algorithms by test suits or not?
Dear Amin Bakhtiari ,
You may use cross validation to validate your model. Cross-validation is a statistical learning method which is used to evaluate and compare the models by partitioning
the data into two portions. One portion of the divided set is used to train or learn the model and the rest of the data is used to validate the model, based on training.
- Badges
- Science topic
- Similar topics
- Computer Science