Get random numbers within one standard deviation
I want to generate random numbers, lets say 100. Now I am using numpy for the purpose like:
print numpy.random.normal(loc=200,scale=50,size=100)
But I want the numbers to be generated only one standard deviation apart from the mean value i.e loc. What would be the best way ?
Should I simply do something like:
print numpy.random.randint(150, high=250, size=100)
or is there any other way of doing it ?
Pick one number
As suggested in @Robert Kern's comment use scipy's truncnorm
from scipy import stats
arr = stats.truncnorm.rvs(-1, 1, loc=200.0, scale=50.0, size=1)
print(arr[0])
The first two parameters indicate the range [a, b] (measured in standard deviations), loc is the mean (the center of the distribution), and scale is the standard deviation (how spread the distribution is).
Plot numbers within -1 to +3 STDs
import matplotlib.pyplot as plt
from scipy import stats
# -1 to +3 standard deviations apart
r = stats.truncnorm.rvs(-1, 3, loc=200.0, scale=50.0, size=10**6)
plt.hist(r, bins=100, color='blue')
plt.xlabel("value")
plt.ylabel("frequency")
plt.show()
Sidenote: randint() and normal() do not pick a number the same way.
Return random integers from the “discrete uniform” distribution in the “half-open” interval.
Draw random samples from a normal (Gaussian) distribution.
The odds of getting any number in the chosen interval using randint() is the same, unlike numbers from a normal distribution (odds of getting a number closer to the peak are greater).
