将阵列标准化为标准正态分布

Question

I think this is a pretty simple question but I wasn't able to find an answer.

I have an array:

array([ 62519, 261500, 1004836, ... , 0, 0])

I would like to convert it to a normal distribution with a min of 0 and a max of 1.

Any suggestions? I was looking at sklearn.preprocess.normalize, but was unable to get it to work for me.

The purpose is that I am creating a scatterplot with numpy, and want to use this third variable to color each point. However, the colors have to be between 0 and 1, and because I have some weird outliers I figured a normal distribution would be a good start.

Let me know if this doesn't make any sense. Thanks & Cheers.

Answer 1

Oh I'm an idiot, I just wanted to standardize it and can just do z = (x- mean)/std . Sorry.

Answer 2

I do not recommend using Standard Normal Distribution for normalization, please consider using frobenius/l2:

Frobenius/2-norm:

Using Numpy:

normalized_z = z / np.linalg.norm(z)

Using pure math lib

normalized_z = z / math.sqrt(max(sum(z**2), 1e-12)) # L2: Matrix Norm

Using Tensorflow

normalized_z = tf.nn.l2_normalize(z,0)