JuniorTheoryCommon

What does this program print?

Python

import numpy as np

train = np.array([[100.0, 1.0], [300.0, 3.0], [100.0, 1.0], [300.0, 3.0]])
live = np.array([[350.0, 4.0], [350.0, 4.0]])
shift = np.abs(live.mean(axis=0) - train.mean(axis=0)) / train.std(axis=0)
print(round(shift[0], 1), round(shift[1], 1))

1.5 2.0

Expecting 1.3 1.7: numpy's std divides by the number of rows, not by rows minus one, so the training spreads are 100 and 1, and the shifts are 1.5 and 2.0.
Expecting 0.75 1.0: the standard deviation is not the range, so values 100 and 300 around the mean 200 give a deviation of 100, not 200.

Which of the two features moved more, judging by raw units and by deviations?
Why is the spread taken from the training set rather than from the new rows?

Back to questions Open topic in roadmap →Answer it yourself — Quiz

Back to questions

JuniorTheoryCommon

What does this program print?

Python

import numpy as np

train = np.array([[100.0, 1.0], [300.0, 3.0], [100.0, 1.0], [300.0, 3.0]])
live = np.array([[350.0, 4.0], [350.0, 4.0]])
shift = np.abs(live.mean(axis=0) - train.mean(axis=0)) / train.std(axis=0)
print(round(shift[0], 1), round(shift[1], 1))

1.5 2.0

Expecting 1.3 1.7: numpy's std divides by the number of rows, not by rows minus one, so the training spreads are 100 and 1, and the shifts are 1.5 and 2.0.
Expecting 0.75 1.0: the standard deviation is not the range, so values 100 and 300 around the mean 200 give a deviation of 100, not 200.

Which of the two features moved more, judging by raw units and by deviations?
Why is the spread taken from the training set rather than from the new rows?

Back to questions Open topic in roadmap →Answer it yourself — Quiz