在使用 keras 模型之前按列标准化数据

Question

I'm working with a large dataset whose data I want to standardize to use with a CNN.

Does keras have a quick utility to standardize a block of numbers column-wise that you can use inside a Sequential model? I'm asking this as i expect eventually the data to be used on-line so ideally this standardization feature could be used on incoming data, ie a trailing moving average of mean and std to normalize the incoming data.

import numpy as np
import pandas as pd

np.random.seed(42)

col_names = ['Column' + str(x+1) for x in range(5)]
training_data = pd.DataFrame(np.random.randint(1,10 **6, 50).reshape(-1,5), columns = col_names)

Answer 1

I am not sure about online, but using sklearn 's StandardScaler() should do the right thing, as described here , seems like the right thing.

Answer 2

We can do from sklearn

from sklearn.preprocessing import StandardScaler
training_data[:]= StandardScaler().fit_transform(training_data.T).T