Corporation AI
The Simplex method from linear algebra looks like it might be well suited to the problem at hand.
Youtube tutorial video on a similar problem space
I need an algorithm that uses past prices, sales and production to adjust future prices and production. It must be subject to changing conditions.
I need an algorithm that uses past prices, sales and production to adjust future prices and production. It must be subject to changing conditions.
No, you don't need an algorithm. You need to understand what data you have, what it is that you want to maximize, subject to what restrictions and using what models for things you need to estimate from incomplete data (e.g. the demand at each price point). Then, and only then, you will need an algorithm to perform the optimization. But the hard work is in the modeling of the problem.
Modeling Random Processes for Engineers and Managers by James Solberg. (I'm assuming probability distributions somewhere in the model.) Easy to understand and lots of problems that are similar to OPs question.
What if I used neural networks? I can train it to try to maximize profit on a minute-to-minute basis. How would I do that? What data would be relevant?
- number of competitor buildings of same type
- number of labourers
- input prices
How would I build it to deal with a variable number of input resources for different production chains? I.e., you might need only stone to produce cement, but for metal you'd need electricity and ore. Would I need to train a neural net for each building type and production chain, if a building type has more than one?