Thank you all for your input, Yes random is what I need!! I checked the docs, and follow your comments, I will use both random.sample and random.shuffle,and random.choice etc. !!
For this particular problem I think ramdom.shuffle() is what I need. I was not very clear or complete in my explanation, and I didn't expect such a discussion on my needs, now I'm embarrassed :) . The list is a data set for training a machine learning algorithm. I want to use 70% of the records (random) for training, but then the remaining 30% are used for validation. This is repeated a few times choose again 70% at random for training and the rest for validation. With random.shuffle() I just iterate for the first 70% of the records for training and I just continue with the remaining 30%. This last 30% we dont care if its random anymore, so shuffle is doing some extra work as it was pointed out. Conceptually something like this: population = range(100) training = random.sample(population,70) validation = set(population).difference(set(training)) But I think this is more costly Thanks Again Mario -- http://mail.python.org/mailman/listinfo/python-list
