Speed Up Your Code with Loop Hoisting : Data Science Code

How to use loop hoisting to speed up your code!

[roygalaasen]

If you want to keep the inefficiency, but solve the correctness issue, you could read from one buffer and write to another, then swap, like double buffer behaviour.

Although I don’t see this as a performance issue at all, because there should not really be a correctness issue to begin with, I like that you pointed it out, as it is a potential trap you could fall for if you are mindlessly programming this thing, or probably mostly with beginners.

[sungchulyonseiackr]

tic = time()
mean = np.mean(myImg, axis=1, keepdims=True)
std = np.std(myImg, axis=1, keepdims=True)
myImgCopy = (myImg - mean) / (std + 1e-6)
toc = ime()
print('Full Vectorization Time : ', toc-tic)

[DefconZero]

Thanks, informative video!


Please be sure to turn off auto-focusing of the camera in the future though.