1
$\begingroup$

Imagine a situation where a business negotiates contracts for the maintenance of widgets it sells.

Situation

  • Customer buys 20 widgets.
  • Customer negotiates contract for widgets to be serviced/replaced.
  • Business makes money by servicing or replacing widgets at a cost lower than the contract payment.
    • (Contract Payment - Actual Maintenance Cost > 0 ... profit!)

Rules

  • Each contract is for multiple widgets.
  • Each widget can only be in use for a specified period of time before repair. However, these repair periods can be stretched for a percentage of time with minimal risk to the widget.
  • Each widget has a limited number of repair cycles before it needs to be replaced.
  • Buying widgets is always more expensive than repairing widgets.
  • Widget inspections must accompany every potential repair.
  • Multiple inspection event types exist (Level 1, Level 2). Level 2 is MUCH more expensive.
    • Level 1 can occur for a small number of widgets rather cheaply.
    • Level 2 must occur for all widgets
  • A repair event is cheaper if it occurs when multiple widgets can be inspected and repaired at once.

What would be appropriate algorithms to forecast the best possible schedule for inspecting/replacing/repairing widgets on a contract over time to maximize profit?

They key goal is to stagger inspection events to have the fewest inspection events and fewest new widget purchases over the life of a contract.

Example - Widget 1 can be in use for 100 hours before repair. Widget 2 can be in use for 110 hours before repair. Stagger Widget 1 forward 10 hours to inspect both at the same time instead of doing an inspection/repair at 100 and 110 hours.

I am doing research on branch and bound ... A* and minimum spanning tree algorithms, but I am not sure which approach was more appropriate for the problem.

Please note that this seems similar to What type of analysis is appropriate for assessing the performance time-series forecasts?

$\endgroup$
1
  • $\begingroup$ I was told this was more easily tackled using simulations, but with limited data science experience, I am wondering if anybody has any resources where I can learn more about different types of simulation techniques for this kind of forecasting. $\endgroup$
    – ds1984
    Commented Mar 29, 2016 at 19:47

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.