Optmize the intial parameters so I can get the best outcome

Hi guys,

I am new to matlab , currently, I have a timeseries dataset and some dependent variables included, and I want to know the best combination of those variables that produce the best outcome.

For example :

the change of offset and setpoint in those two parameters would modify the outcome, And I want to know what the best result can be if I tune the offset point and setpoint.

What sort of toolbox do I need to solve this problem?

Many thanks

[Torsten]

Optimization Toolbox, I guess.

If you could explain your problem in more detail, maybe we could advice a suitable solver from the Toolbox.

[John Li]

Hi thanks for your reply , yes I have looked the opmtization toolbox but you need a formula for that right? and it is for linear function.

what I have is a timesereis dataset from a machine and there are lots of external factor impact its result, I want to understand if I change its parameters ( offset and setpoint) what sort of results ( another time series chart) would I get, sort of a preventative prediction.

[Chunru]

For all optimization problem, you need to define an object function. For your case, the objective function is f(offset, setpoint) which you have to specify according to the problem you want to solve. If offset and setpoint are scalar, this will result in a 2D optimization problem. You do need to specify that the goodness of the results (you mentioned it is a time series) with a given offset and setpoint.

For a 1D/2D optimization, the brute force searching may be possible. Otherwise, you may need optimization toolbox for more suitable approaches.

Answer 1

[Varun]

Hi John,

Looks like you have a timeseries dataset and some dependent variables included, and you want to know the best combination of those variables that produce the best outcome. You also want to know the toolboxes which can solve this problem.

To solve the problem of finding the best combination of variables that produce the best outcome, you can use optimization techniques. MATLAB provides several toolboxes that can help you with this task. Here are two commonly used toolboxes for optimization:

Optimization Toolbox: This toolbox provides a comprehensive set of functions for solving optimization problems, including both linear and nonlinear programming. It offers various optimization algorithms and techniques such as constrained and unconstrained optimization, global optimization, and multi-objective optimization. You can use this toolbox to define an objective function that evaluates the outcome based on the variables, and then optimize the variables to find the best combination.

Global Optimization Toolbox: This toolbox focuses on global optimization problems, which involve finding the global minimum or maximum of a function within a given range. It offers algorithms specifically designed for global optimization, including genetic algorithms, particle swarm optimization, and simulated annealing. If your problem involves finding the global best outcome among various combinations, this toolbox can be useful.

Both toolboxes provide functions and algorithms that can be adapted to your specific problem. You can define your objective function based on the variables and the desired outcome, and then use the optimization functions to search for the best combination of variables.

To learn more about using these toolboxes, please refer to the following documentations:

Optimization Toolbox: https://www.mathworks.com/help/optim/index.html

Global Optimization Toolbox: https://www.mathworks.com/help/gads/index.html

Hope it helps.