0

I have a few queries regarding my implementation of the Genetic Algorithm (GA).

I have a lot of parameters in which I have to find the best combination of these parameters to maximize the value of the function. Hence, I am using GA. The maximizing or fitness function mentioned above are some complex equations that run over a couple of data points, and the configuration from the population is applied to these equations. Now, the GA tries to maximize the output of this function which is the average of each data point (each data point gives us a value). Now assuming that I have fewer data points and the data points are diverse -

  1. Is GA the right approach (considering there is no lack of data)?
  2. If lack of Data is considered, is the GA the best approach?
  3. Are there ways to determine the ideal amount of data required for GA to work correctly in my scenario mentioned above? I have around 100 data points in my case.
  4. So we have prototyped it and run it, but it seems to be giving random results; any ways to improve it?
ashwin ram kumar
  • 1
  • 1
    Can you clarify what problem you're trying to solve? You're looking to maximize some function $f(x)$. What is your GA evolving? Is it the values of $x_i$ that maximize $f$ and those $x_i$ values are your $n$ data points? – deong Jun 13 '22 at 13:14
  • Regarding your 4th question, since a GA is a stochastic method the results will always be somewhat random. You can reduce the "randomness" by increasing the population size, increasing the number of generations, and also combining the results of several independent evolutions. Making those changes essentially ensures you are exploring more of your search space and increases your chance of finding a global optima. Regarding your other questions, I think you need to provide some more context to allow for an appropriate answer. – Nate Jun 13 '22 at 19:02

0 Answers0