Why random generation matters in NumPy - Challenge Your Understanding

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of numpy random seed effect

What will be the output of the following code snippet?

import numpy as np
np.random.seed(0)
sample1 = np.random.randint(1, 10, 3)
np.random.seed(0)
sample2 = np.random.randint(1, 10, 3)
print(sample1)
print(sample2)

NumPy

import numpy as np
np.random.seed(0)
sample1 = np.random.randint(1, 10, 3)
np.random.seed(0)
sample2 = np.random.randint(1, 10, 3)
print(sample1)
print(sample2)

A

[6 1 4]
[6 1 4]

B

[6 1 4]
[2 8 8]

C

[1 6 4]
[1 6 4]

D

[2 8 8]
[2 8 8]

💡 Hint

Setting the seed resets the random number generator to the same starting point.

Using np.random.seed(0) before generating random numbers ensures the same sequence is produced each time. So sample1 and sample2 are identical.

🧠 Conceptual

intermediate

Why set a random seed in data science?

Why is it important to set a random seed when generating random data in data science projects?

A To ensure the random numbers are different every time the code runs

B To make the random number generation reproducible for debugging and sharing results

C To speed up the random number generation process

D To increase the randomness quality of the generated numbers

💡 Hint

Think about sharing your work with others or rerunning your code later.

Setting a random seed makes the random number generation reproducible, so results can be repeated exactly, which is important for debugging and sharing.

❓ data_output

advanced

Effect of different seeds on random samples

What is the shape and content type of the output from this code?

import numpy as np
np.random.seed(42)
sample = np.random.normal(loc=0, scale=1, size=(2,3))
print(type(sample))
print(sample.shape)

NumPy

import numpy as np
np.random.seed(42)
sample = np.random.normal(loc=0, scale=1, size=(2,3))
print(type(sample))
print(sample.shape)

A

<class 'numpy.ndarray'>
(2, 3)

B

<class 'list'>
(2, 3)

C

<class 'numpy.ndarray'>
(3, 2)

D

<class 'list'>
(3, 2)

💡 Hint

Check the type returned by np.random.normal and the shape argument.

The function np.random.normal returns a numpy array with the shape specified by the size parameter, here (2, 3).

🔧 Debug

advanced

Identify the error in random number generation

What error will this code produce?

import numpy as np
np.random.seed('seed')
sample = np.random.randint(0, 5, 4)
print(sample)

NumPy

import numpy as np
np.random.seed('seed')
sample = np.random.randint(0, 5, 4)
print(sample)

A SyntaxError: invalid syntax

B ValueError: seed must be between 0 and 2**32 - 1

C No error, prints 4 random integers between 0 and 4

D TypeError: 'str' object cannot be interpreted as an integer

💡 Hint

Check the type of the argument passed to np.random.seed.

The seed must be an integer or array of integers. Passing a string causes a TypeError.

🚀 Application

expert

Using random generation for train-test split reproducibility

You want to split your dataset into training and testing parts randomly but reproducibly. Which code snippet correctly achieves this?

A

import numpy as np
indices = np.random.permutation(10)
train_idx = indices[:7]
test_idx = indices[7:]

B

import numpy as np
np.random.seed('123')
indices = np.random.permutation(10)
train_idx = indices[:7]
test_idx = indices[7:]

C

import numpy as np
np.random.seed(123)
indices = np.random.permutation(10)
train_idx = indices[:7]
test_idx = indices[7:]

D

import numpy as np
np.random.seed(123)
indices = np.random.randint(0, 10, 10)
train_idx = indices[:7]
test_idx = indices[7:]

💡 Hint

Think about reproducibility and correct use of random permutation for splitting.

Setting the seed ensures reproducibility. Using np.random.permutation shuffles indices without repeats. Option C correctly sets the seed and uses permutation.