[open] Cumultative Chance for a random event

Comments

• sebastiaan

  June 2013 edited June 2013

  Hi Henrik,

  Sure, with a bit of scripting this type of thing can be implemented quite easily. However, it seems that you're describing two different scenarios:

  • The probability of an event is contingent on whether the event occurred before. (First example.)
  • The probability of an event is contingent on how often another (non-random) event has occurred before. (Traffic-jam example.)

  What exactly do you want to do?

  Cheers,
  Sebastiaan

  
  
  Check out SigmundAI.eu for our OpenSesame AI assistant!
  
• Henrik

  June 2013 edited 10:14PM

  Hi sebastiaan,

  Oh I see the problem, my bad. I want the first example. A random event that occurs with increasing chances depending on wether or not it has occured before.

  Best regards,

  Henrik
• sebastiaan

  June 2013 edited 10:14PM

  Ah, I see. Basically, you will want to maintain a variable, let's called it [prob], that specifies the probability that an event occurs. Each time that you determine (randomly) whether the event actually occurs, [prob] decreases when it does, and increases when it doesn't, as in the script below:

  from random import random
  
  # random() returns a number between 0 and 1
  # so this sets `event_occurs` to True or False
  # with a probability of [prob].
  prob = self.get('prob')
  event_occurs = random() < prob
  
  # If the event does not occur, increase the
  # probability for the next time, but not
  # higher than 1.
  if not event_occurs:
      prob += .1
      if prob > 1:
          prob = 1
  # If the event does occur, decrease the probability,
  # but not below 0.
  else:
      prob -= .1
      if prob < 0:
          prob = 0
  
  # Store [prob] and [event_occurs]
  exp.set('prob', prob)
  exp.set('event_occurs', event_occurs) # Will be set to 'yes' or 'no'
  

  This script assumes that [prob] has some kind of starting value. And, of course, it depends on the experiment what the logic will be, and how the variable [event_occurs] is used. But in general, this should get you started.

  Here you see the script in the context of a working dummy experiment:

  • http://pastebin.com/01bdJFy1

  Cheers!
  Sebastiaan

  
  
  Check out SigmundAI.eu for our OpenSesame AI assistant!