This code is to randomize an array with a specific proportion
a0 = np.full((17),0)
a1 = np.full((30),1)
a2 = np.full((38),2)
a3 = np.full((15),3)
options = np.concatenate((a0,a1,a2,a3))
np.random.shuffle(options)
it works as expected generating [a0, a1, a2, a3] with the specific proportion [17%, 30%, 38%, 15%]
However, it becomes ugly when the number of "ax" grow large, like [a0, a1, ... , a999] with [1%, 5%, ... , .2%]
Does Python or NumPy have API to do that work?
You can use a dict where the key is number and the value is the repeats. Then use a list comprehension to build the list to concatenate.
d = {
0: 17,
1: 30,
2: 38,
3: 15
}
options = no.concatenate([np.full((v), k) for k, v in d.items()])
So if you grow the basenumber the individual percentage decreases, if you want to hold the same percentages you have to increment the size of the arrays with the same proportion. Instead of writing '17' you could try to write 0.17 * len(total array), if this does help
