It is often the case that we need to generate a random number or an array of random numbers. Luckily, NumPy has a module named random specifically for this purpose.

Note

random generates pseudo-random numbers, not truly random numbers. However, for most purposes, this is more than enough.

The two most commonly used functions of the random module are:

• rand();
• randint().

rand()

The numpy.random.rand() function is used to generate either a random float number or an array of random floats from a uniform distribution over [0, 1).

Its only possible arguments are the dimensions of the array. If no argument is passed, rand() generates a random float number (scalar).

Let’s clarify with an example:

First, we generated a random float number. Next, we generated a random 1D array with 5 elements by setting its size to 5. Finally, we generated a random 2D array of shape 3x4.

Note

Dimensions in the rand() function should be specified as separate integer parameters, not a tuple of integers.

randint()

The numpy.random.randint function is used to generate either a random integer or an array of random integers from a discrete uniform distribution within a specified interval.

Its three most important parameters are low (the only required parameter), high, and size. The interval is [low, high) (from low inclusive to high exclusive). However, if high is not specified, then the interval is [0, low).

Let’s look at an example:

There is actually nothing complicated here; everything is rather straightforward.

Note

Unlike rand(), we specify the dimensions of the array via a single parameter size, passing either an integer or a tuple of integers.

As always, here is the documentation for you: numpy.random.rand and numpy.random.randint.