We have sufficient behavioral data of a real time process (e.g., task scheduling in a cloud) to construct its model. Which way is better for finding optimum solutions for this process, using data analysis and prediction methods or meta-heuristic algorithm?
You surely can use predictive methods. Please take a look into Applied Predictive Modeling by Max Kuhn, Springer, where such a study-case is presented
If I got your question correctly, there is no way in current technology to find the optimal solution for combinatorial problems in real-time, particularly NP-complete problems, and this applies to the scheduling problem since it is NP-complete problem.
well, near-optimal and approximate solutions are available in real-time for most of these problems.using GA, particle swarm optimization, differential evolution, etc.
- Badges
- Science topic