Optimisation
Heuristic algorithms
Approaches that produce good solutions in acceptable time for portfolio, allocation and scheduling problems where exact methods are too costly.
What we do here
For some problems, searching for the best answer costs more than the answer is worth. Portfolio allocation, resource assignment and scheduling stop being solvable by exact methods in reasonable time as the option space grows. Heuristics step in here: they do not guarantee the optimum, they produce a good enough result in acceptable time.
Scope
- Modelling the problem as constraints and an objective function
- Selecting the heuristic or metaheuristic that fits the problem
- Tuning the trade-off between solution quality and run time
- Comparing results against an exact method or the current practice
- Repeatable execution in the production environment
How we work
A heuristic is worth exactly as much as the baseline it is compared against. Every engagement starts by measuring what the current approach produces; the new method is then evaluated against that baseline on the institution's own data. No comparison, no claim of improvement.
What are you trying to solve?
Tell us your system requirements and our team will scope it with you.