Random numbers

Using the random module, we can generate pseudo-random numbers. The function random() generates a random number between zero and one [0, 0.1 .. 1].  Numbers generated with this module are not truly random but they are enough random for most purposes.

Random number between 0 and 1.
We can generate a (pseudo) random floating point number with this small code:

from random import *
print random()     # Generate a pseudo-random number between 0 and 1.

Generate a random number between 1 and 100
To generate a whole number (integer) between one and one hundred use:

from random import *
print randint(1, 100)    # Pick a random number between 1 and 100.

This will print a random integer. If you want to store it in a variable you can use:

from random import *
x = randint(1, 100)    # Pick a random number between 1 and 100.
print x

Random number between 1 and 10
To generate a random floating point number between 1 and 10 you can use the uniform() function

from random import *
print uniform(1, 10)

Picking a random item from a list

Fun with lists
We can shuffle a list with this code:

from random import *
items = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
print items

To pick a random number from a list:

from random import *
items = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
x = sample(items,  1)   # Pick a random item from the list
print x[0]
y = sample(items, 4)    # Pick 4 random items from the list
print y

We can do the same thing with a list of strings:

from random import *
items = ['Alissa','Alice','Marco','Melissa','Sandra','Steve']
x = sample(items,  1)   # Pick a random item from the list
print x[0]
y = sample(items, 4)    # Pick 4 random items from the list
print y

17 thoughts on “Random numbers

  1. Gauge - May 5, 2016

    Just a quick question about sample() — when used to fill values in an equation it returns:

    TypeError: can’t multiply sequence by non-int of type ‘float’

    I’m curious why this isn’t allowed, and if there is an alternative that I haven’t been able to find for extracting multiple random values from a series of value. For now, all I’ve come up with is using the choice() alternative with a loop for however large my sample pool needs to be, but I have a feeling there’s a better way to do it. Admittedly, I haven’t read through all of the documentation for random thoroughly, but I haven’t found a better solution. I figured that someone with a bit of knowledge would be my best bet. Any suggestions? As an aside, I should note I’m using 2.7 to learn with. I know they changed some stuff with Python 3.x, so I’m not sure if this is even an issue with newer revisions.

    1. Frank - May 5, 2016

      Are you multiplying with a sequence, perhaps something like this?:

      seq = [1,2,3,4,5]
      y = seq * randomElement

      To do so, we have to iterate over every element:

      from random import *
      items = [1, 2.13, 3.24, 4.25, 5.46, 6.57, 7.68, 8.79, 9.810, 10.911]
      seq = [1,2,3,4,5]
      # Get a list of 1 random numbers, convert to one random number.
      randomElement = sample(items,  1)
      randomElement = randomElement[0]
      # Multiple every element of the list
      for element in seq:
          y = element * randomElement
          print("y=", y)

      Alternative is to use one of the two methods below:

      from random import *
      seq = [1,2,3,4,5]
      # Get a list of 1 random numbers, convert to one random number.
      randomElement = sample(seq,  1)
      randomElement = randomElement[0]
      # Multiple every element of the list, using list comprehension
      l = [randomElement*x for x in seq]
      # Multiple every element using map
      l = list(map(lambda x:randomElement*x, seq))

      I hope that helps. If not, could you post the code?

      1. Gauge - May 5, 2016

        I think you answered my question pretty well. My mistake was in not addressing the derived sample with an index value, so I was essentially attempting mathematics against a list rather than single values from within the list. It somehow slipped past me that sample returned a list. I guess it’s a beginner mistake and lack of attention on my part.

        Quick questions, though:

        Is the behavior of a loop iteration of choice() essentially the same as the returned sample() list? For clarification, a simple example would be a loop that iterates 4 times using choice() and storing the returned values in a list, versus a sample() of 4 the same values. Would that essentially return the same chance of random values?

        Also, supposing they essentially return the similar randomness in their values, is there a performance difference over larger iterations? Common sense tells me that since the looping choice() jumps right to the chase, but a sample() first stores the values in a new list and that list then needs to be iterated over, looping choice() would be the cleaner and more efficient alternative. Am I mistaken?

        Thanks for the quick response, I appreciate your answer and it definitely helped me understand where I originally went wrong.

        1. Gauge - May 6, 2016

          Apologies for the double response. The original response didn’t exist on a new browser, and the original browser was stuck on the spinning dots for the comment section, so I assumed it hadn’t gone through the first time. Feel free to delete whichever one you think is best.

          1. Frank - May 6, 2016

            No problem.

            Practically that would do the same. There is a slight difference, choice() will raise an IndexError if the sequence is empty. Sample will throw a ValueError. This matters if you implement error handling (try/catch statement). Performance wise there is a big difference.

            Performance measuring
            These are the functions inside the Python code:
            (however depending on your version of Python they may have a different implementation)

                def choice(self, seq):
                    """Choose a random element from a non-empty sequence."""
                        i = self._randbelow(len(seq))
                    except ValueError:
                        raise IndexError('Cannot choose from an empty sequence')
                    return seq[i]
                def sample(self, population, k):
                    if isinstance(population, _collections.Set):
                        population = tuple(population)
                    if not isinstance(population, _collections.Sequence):
                        raise TypeError("Population must be a sequence or Set.  For dicts, use list(d).")
                    randbelow = self._randbelow
                    n = len(population)
                    if not 0 <= k <= n:
                        raise ValueError("Sample larger than population")
                    result = [None] * k
                    setsize = 21        # size of a small set minus size of an empty list
                    if k > 5:
                        setsize += 4 ** _ceil(_log(k * 3, 4)) # table size for big sets
                    if n <= setsize:
                        # An n-length list is smaller than a k-length set
                        pool = list(population)
                        for i in range(k):         # invariant:  non-selected at [0,n-i)
                            j = randbelow(n-i)
                            result[i] = pool[j]
                            pool[j] = pool[n-i-1]   # move non-selected item into vacancy
                        selected = set()
                        selected_add = selected.add
                        for i in range(k):
                            j = randbelow(n)
                            while j in selected:
                                j = randbelow(n)
                            result[i] = population[j]
                    return result

            Notice that the sample() function executes a lot more operations. My gut feeling tells that the choice() function would be smaller, but let’s put that to the test.

            I came up with this program:

            from random import *
            import time
            def testSample(size):
                start_time = time.time()
                seq = [1,2,3,4,5]
                for i in range(0,size):
                    randomElements = sample(seq, 1)[0]
                print("--- %s seconds ---" % (time.time() - start_time))
            def testChoice(size):
                start_time = time.time()
                seq = [1,2,3,4,5]
                for i in range(0,size):
                    randomElement = choice(seq)
                print("--- %s seconds ---" % (time.time() - start_time))

            Output with Python 2.7:
            — 5.25197696686 seconds —
            — 1.08564114571 seconds —

            Output with Python 3.4:
            — 17.56459665298462 seconds —
            — 2.1325480937957764 seconds —

            I would say you are correct, choice() is faster.

            In terms of randomness, all of the numbers generated by the random module as pseudo-random. Python 2.3 uses the Wichmann-Hill algorithm , later versions use the MersenneTwister algorithm (Python implementation below)

        2. Gauge - May 6, 2016

          I’m assuming your response hit the maximum depth, so I’ll respond to myself here for the sake of keeping it together…

          I appreciate your response, and I’m actually relieved that my line of thought seems to be heading in the right direction. I just had a finishing question for clarity.

          If I’m understanding you right, each time a number is selected, no matter which version of “random” you use, it runs the same algorithm at the core to return each value with additional modification for specific behavior. Meaning, the returned values essentially exist in the same frequencies of pseudo-randomness? And,

          1. Frank - May 7, 2016

            Yes, you are understanding right. The functions choice() and sample() use the same algorithm, which is implemented in the method random(). It is called indirectly from the method randbelow().

            Python 2.7 would simply call the algorithm (random()) directly:

               def choice(self, seq):
                    return seq[int(self.random() * len(seq))]  # raises IndexError if seq is empty
               def sample(self,population,k):
                    j = _int(random() * (n-i))

            Is there a question after “And,” ? It seems to have gone missing.