Clever Geek Handbook

Using random.choice () with preference and uniqueness

I have a list:

decisions = ['yes', 'no', 'unknown']

I am writing to a file using this list:

for x in range(0, 100):
    file.write(random.choice(decisions))

What would be the most efficient way to ensure that 70% of the written values ​​were something like "unknown"?

I want randomness to some extent, but also want the 70 values ​​written to the file to be of a specific type. I plan on getting this percentage from the user so that he can change every run.

If I had another list that was much larger and would like to enforce uniqueness (no duplicate values, but also randomly ordered), what would be the best method?

+3


source to share


4 answers


If you are using NumPy for you, this is pretty straightforward to implement:

np.random.choice(['yes', 'no', 'unknown'], p=[0.15, 0.15, 0.7])

The second array is p

added to one and represents the probability that the corresponding entry in the first array will be selected.

The above will ensure that the "unknown" is chosen every time with a 70% probability.

The docs are here .



If you want 100 options with exactly 70 "unknown" entries and 30 "yes" or "no":

hundred_choices = ['unknown']*70 + [random.choice(['yes', 'no']) for _ in range(30)]

... and then shuffle hundred_choices

with random.shuffle

.

+6


source


To get exactly a certain percentage of the values ​​written, you can use the shuffle () function, which contains the number of items you need.



choices = ['unknown']*70 + ['yes']*15 + ['no']*15
random.shuffle(choices)
+4


source


If you want 70% chance for 'unknown'

:

decisions = 70 * ['unknown'] + 15 * ['yes'] + 15 * ['no']
choice(decisions)
0


source


If you want exactly 70% 'unknown'

s:

from random import choice, shuffle

values = 70 * ['unknown']                # your 70 unknowns
for i in range(30):
    values.append(choice(['yes', 'no'])) # fill the rest randomly
shuffle(values)                          # shuffle

for x in values:
    file.write('{}\n', x)
0


source






More articles:


All Articles