Random Multinomial

np.random.multinomial(n, [p_0, p_1, p_2, ...])

Determines the number of times each outcome occurs based on the given probabilities.

Input:

n : integer

Number of random samples to draw.

[p_0, p_1, p_2, ...] : list or array of floats

Probabilities for each outcome.

Returns:

Number of times each outcome is drawn

Return Type:

array of integers

Note:

• Each element in [p_0, p_1, p_2, ...] must be between 0 and 1. The elements in [p_0, p_1, p_2, ...] must sum to 1.

---

np.random.multinomial(100, [0.2, 0.5, 0.3])

array([19, 54, 27])

np.random.multinomial(100, [0.1, 0.1, 0.8]) 

array([12, 8, 80])

#use np.random.choice to achieve the function of np.random.multinomial
def manual_multinomial(n, p):
    values = np.arange(len(p))
    choices = np.random.choice(values, size=n, replace=True, p=p)
    value_counts = np.array([])
    for value in values:
        value_count = np.count_nonzero(choices == value)
        value_counts = np.append(value_counts, value_count)
    return value_counts.astype(int)

manual_multinomial(100, [0.1, 0.1, 0.8])

array([10, 16, 74])