First watch/learn/try:

• The Conference scheduling video's https://www.optaplanner.org/learn/useCases/conferenceScheduling.html
• Then solve a conference scheduling dataset on your machine
• Then open conference scheduling dataset and change the weights in the first tabs and solve that.
  • Make sure there is never a track conflict. Verify that in the resulting spreadsheet.
• Now watch the examination video. Look for it in optaplanner.org -> learn -> video's.
  • Notice how it too, plays with constraint weights to for example increase the importance of "mixed duration" constraint
  • Try that yourself too
• In the optaplanner docs, read the entire first section in the chapter on score calculation chapter 5, all the way up to "pareto" optimization. Understand the difference between weights and levels. Also see my devoxx morocco recording to see the score effects in action.
• In the optaplanner docs, read about ConstraintConfiguration (ignore the code, focus on the business principles behind it)

Once this is done, you'll have a good idea what constraint tweaking is. Then it's time to understand the need for doing simulations:

• What if I increase the importance of "mixed duration"? What is the impact? How much will the other constraints suffer?

Once that is understood too, design a screen for optaweb-employee-rostering that has the same data as the first sheet in the conference scheduling spreadsheet, but a lot more user friendly.

Ps: Ask Christopher a link to the latest optaweb-employee rostering on openshift.