Setting random seed for reproducibility in NumPy - Cheat Sheet & Quick Revision

Recall & Review

beginner

What does setting a random seed in numpy do?

Setting a random seed makes sure that the random numbers generated are the same every time you run the code. This helps in getting reproducible results.

beginner

How do you set a random seed in numpy?

You use np.random.seed(your_number). Replace your_number with any integer to fix the sequence of random numbers.

beginner

Why is reproducibility important in data science?

Reproducibility means you or others can run the same code and get the same results. This is important for checking work, sharing findings, and debugging.

beginner

What happens if you don't set a random seed before generating random numbers?

If you don't set a seed, numpy will generate different random numbers each time you run the code, which can make results vary and be hard to reproduce.

beginner

Example: How to generate 5 random numbers between 0 and 1 with a fixed seed?

```python
import numpy as np
np.random.seed(42)
rands = np.random.rand(5)
print(rands)
```
This will always print the same 5 random numbers.

What function sets the random seed in numpy?

A numpy.set_seed()

B numpy.random.set()

C numpy.random.seed()

D numpy.seed.random()

The correct function to set the seed is numpy.random.seed().

Why should you set a random seed in your code?

A To avoid using random numbers

B To make random numbers different every time

C To speed up random number generation

D To make random numbers reproducible

Setting a seed makes random numbers reproducible, so results stay the same each run.

What type of value do you pass to numpy.random.seed()?

A An integer

B A float

C A string

D A boolean

You pass an integer to numpy.random.seed() to fix the random sequence.

If you run numpy.random.rand(3) twice without setting a seed, what happens?

A You get different 3 numbers each time

B You get zeros

C You get an error

D You get the same 3 numbers both times

Without a seed, random numbers differ each run.

Which of these is a benefit of reproducibility in data science?

A Code runs faster

B Results can be trusted and checked

C Random numbers are more random

D You don't need to write comments

Reproducibility helps others trust and verify your results.

Explain how to set a random seed in numpy and why it is useful.

Think about how to get the same random numbers every time.

Describe what happens if you do not set a random seed before generating random numbers in numpy.

Consider the effect on results when running code multiple times.

Practice

1. What is the main purpose of setting a random seed in numpy?

easy

A. To increase the range of random numbers

B. To speed up random number generation

C. To generate only positive random numbers

D. To make random number results repeatable

Solution

1. Step 1: Understand what a random seed does

   Setting a random seed fixes the starting point for random number generation.

2. Step 2: Effect of fixed seed on results

   This means the same sequence of random numbers is produced every time the code runs.

3. Final Answer:

   To make random number results repeatable -> Option D

4. Quick Check:

   Random seed = repeatable results [OK]

Hint: Seed fixes randomness to get same results every run [OK]

Common Mistakes:

• Thinking seed speeds up random generation
• Believing seed changes number range
• Assuming seed makes numbers positive only

2. Which of the following is the correct syntax to set a random seed to 42 in numpy?

easy

A. np.random.seed(42)

B. np.seed.random(42)

C. np.random.set_seed(42)

D. np.random.seed = 42

Solution

1. Step 1: Recall the correct function name and usage

   The correct function to set seed in numpy is np.random.seed().

2. Step 2: Check the argument and syntax

   The seed value is passed as an integer inside the parentheses, like np.random.seed(42).

3. Final Answer:

   np.random.seed(42) -> Option A

4. Quick Check:

   Correct function call = np.random.seed(42) [OK]

Hint: Use np.random.seed(number) exactly [OK]

Common Mistakes:

• Swapping function and module names
• Using incorrect function like set_seed
• Assigning seed instead of calling function

3. What will be the output of the following code?

import numpy as np
np.random.seed(0)
print(np.random.randint(1, 10))
np.random.seed(0)
print(np.random.randint(1, 10))

medium

A. Different numbers both times

B. 5 followed by different number

C. 5 followed by 5

D. Error due to repeated seed setting

Solution

1. Step 1: Understand effect of setting seed before each random call

   Setting the seed to 0 resets the random number generator to the same start point.

2. Step 2: Predict output of np.random.randint(1, 10) after resetting seed

   Both calls produce the same random integer because the seed is reset before each call.

3. Final Answer:

   5 followed by 5 -> Option C

4. Quick Check:

   Same seed = same random number [OK]

Hint: Resetting seed repeats same random number [OK]

Common Mistakes:

• Assuming different numbers after resetting seed
• Thinking repeated seed causes error
• Ignoring seed reset effect

4. Identify the error in this code snippet that tries to set a random seed:

import numpy as np
np.random.seed = 123
print(np.random.rand())

medium

A. Missing import statement

B. Seed is assigned instead of called as a function

C. Seed value must be a float, not integer

D. np.random.rand() is incorrect function

Solution

1. Step 1: Check how seed is set

   The code assigns 123 to np.random.seed instead of calling it as a function.

2. Step 2: Understand consequence of assignment

   This overwrites the seed function with an integer, causing errors on later random calls.

3. Final Answer:

   Seed is assigned instead of called as a function -> Option B

4. Quick Check:

   Use np.random.seed(123), not assignment [OK]

Hint: Call seed() as function, don't assign it [OK]

Common Mistakes:

• Assigning seed instead of calling
• Using wrong random function
• Ignoring import statement

5. You want to generate a reproducible array of 3 random floats between 0 and 1 using numpy. Which code snippet achieves this correctly?

hard

A. np.random.seed(7)
arr = np.random.rand(3)
print(arr)

B. np.random.seed = 7
arr = np.random.rand(3)
print(arr)

C. np.random.set_seed(7)
arr = np.random.rand(3)
print(arr)

D. np.random.seed(7)
arr = np.random.randint(3)
print(arr)

Solution

1. Step 1: Set the random seed correctly

   Use np.random.seed(7) to fix the random sequence.

2. Step 2: Generate 3 random floats between 0 and 1

   np.random.rand(3) creates an array of 3 floats in [0,1).

3. Final Answer:

   np.random.seed(7) arr = np.random.rand(3) print(arr) -> Option A

4. Quick Check:

   Correct seed + rand(3) = reproducible floats [OK]

Hint: Seed with np.random.seed(), use np.random.rand(3) [OK]

Common Mistakes:

• Assigning seed instead of calling
• Using non-existent set_seed function
• Using randint instead of rand for floats