Correlation coefficient with np.corrcoef() in NumPy - Cheat Sheet & Quick Revision

Recall & Review

beginner

What does the correlation coefficient measure?

It measures the strength and direction of a linear relationship between two variables, ranging from -1 (perfect negative) to 1 (perfect positive).

beginner

What is the purpose of np.corrcoef() in numpy?

It calculates the correlation coefficient matrix between input arrays, showing how variables relate linearly.

beginner

How do you interpret a correlation coefficient of 0?

It means there is no linear relationship between the two variables, but they might still have other types of relationships.

intermediate

What shape of output does np.corrcoef() return when given two 1D arrays?

It returns a 2x2 matrix where diagonal elements are 1 (correlation of each variable with itself) and off-diagonal elements are the correlation coefficients between the two variables.

intermediate

Why is it important to check correlation before building a predictive model?

Because strong correlations can indicate important relationships between variables, helping to select features and understand data patterns.

What is the range of values for the correlation coefficient?

A -1 to 1

B 0 to 1

C -100 to 100

D 0 to infinity

The correlation coefficient ranges from -1 (perfect negative) to 1 (perfect positive).

What does np.corrcoef(x, y) return when x and y are 1D arrays?

A A list of correlation values

B A single correlation value

C A 2x2 matrix of correlation coefficients

D An error

It returns a 2x2 matrix showing correlations between x and y and themselves.

If the correlation coefficient between two variables is close to 0, what does it mean?

A No linear relationship

B Strong positive relationship

C Strong negative relationship

D Perfect correlation

A value near 0 means no linear relationship between variables.

Which numpy function is used to calculate correlation coefficients?

A np.cov()

B np.corrcoef()

C np.mean()

D np.std()

np.corrcoef() calculates correlation coefficients.

What does a correlation coefficient of -1 indicate?

A Random relationship

B No relationship

C Perfect positive linear relationship

D Perfect negative linear relationship

A value of -1 means a perfect negative linear relationship.

Explain how to use np.corrcoef() to find the correlation between two data arrays.

Think about what the matrix elements represent.

Describe what the correlation coefficient tells us about the relationship between two variables.

Consider how close the value is to 1 or -1.

Practice

1. What does the function np.corrcoef() calculate in numpy?

easy

A. The correlation coefficient between two or more data sets

B. The sum of all elements in an array

C. The mean value of a data set

D. The standard deviation of a data set

Solution

1. Step 1: Understand the purpose of np.corrcoef()

   This function calculates how two or more data sets move together, which is called the correlation coefficient.

2. Step 2: Identify the correct description

   Among the options, only The correlation coefficient between two or more data sets describes the correlation coefficient calculation.

3. Final Answer:

   The correlation coefficient between two or more data sets -> Option A

4. Quick Check:

   np.corrcoef() = correlation coefficient [OK]

Hint: Correlation means how two sets move together [OK]

Common Mistakes:

• Confusing correlation with sum or mean
• Thinking it calculates standard deviation
• Assuming it returns a single number always

2. Which of the following is the correct syntax to calculate the correlation coefficient matrix between arrays a and b using numpy?

easy

A. np.corrcoef(a, b)

B. np.corrcoef[a, b]

C. np.corrcoef(a & b)

D. np.corrcoef(a + b)

Solution

1. Step 1: Recall the function call syntax

   Functions in numpy use parentheses () to pass arguments, so np.corrcoef(a, b) is correct.

2. Step 2: Identify incorrect syntax

   Options B uses square brackets which is invalid for function calls; C uses bitwise AND which is wrong; D adds arrays which changes data meaning.

3. Final Answer:

   np.corrcoef(a, b) -> Option A

4. Quick Check:

   Use parentheses for function calls [OK]

Hint: Use parentheses () to call functions in numpy [OK]

Common Mistakes:

• Using square brackets instead of parentheses
• Using operators inside function arguments incorrectly
• Passing combined arrays instead of separate arguments

3. What is the output of the following code?

import numpy as np
x = np.array([1, 2, 3])
y = np.array([4, 5, 6])
result = np.corrcoef(x, y)
print(result)

medium

A. [[0. 1.]
[1. 0.]]

B. [[1. 1.]
[1. 1.]]

C. [[1. 0.]
[0. 1.]]

D. [[1. -1.]
[-1. 1.]]

Solution

1. Step 1: Calculate correlation between x and y

   Arrays x and y increase together perfectly, so their correlation coefficient is 1.

2. Step 2: Understand np.corrcoef output matrix

   The diagonal is correlation of each array with itself (1), off-diagonal is correlation between x and y (1).

3. Final Answer:

   [[1. 1.] [1. 1.]] is incorrect because the matrix is symmetric with 1 on diagonal and off-diagonal representing correlation between x and y.

4. Correction:

   The actual output is

   [[1. 1.]
    [1. 1.]]

   which matches [[1. 1.] [1. 1.]], but numpy returns correlation matrix with 1 on diagonals and correlation coefficient off-diagonal.

5. Quick Check:

   Perfect positive correlation = 1 [OK]

Hint: Perfectly increasing arrays have correlation 1 [OK]

Common Mistakes:

• Confusing correlation with covariance
• Misreading the matrix layout
• Expecting zeros off-diagonal for correlated arrays

4. The code below is intended to calculate the correlation coefficient between two arrays a and b. What is the error?

import numpy as np
a = np.array([1, 2, 3])
b = np.array([4, 5, 6])
result = np.corrcoef(a & b)
print(result)

medium

A. np.corrcoef() cannot handle numpy arrays

B. Using bitwise AND (&) instead of passing two arrays separately

C. Arrays a and b have different lengths

D. Missing import statement for numpy

Solution

1. Step 1: Identify the incorrect operator usage

   The code uses bitwise AND (&) between arrays, which produces a new array, not two separate arrays.

2. Step 2: Correct usage of np.corrcoef()

   np.corrcoef() expects separate arrays as arguments, like np.corrcoef(a, b), not combined with &.

3. Final Answer:

   Using bitwise AND (&) instead of passing two arrays separately -> Option B

4. Quick Check:

   Pass arrays separately, no & operator [OK]

Hint: Pass arrays as separate arguments, not combined with & [OK]

Common Mistakes:

• Using & instead of comma to separate arrays
• Assuming & merges arrays correctly
• Ignoring error messages about argument types

5. Given two arrays representing daily temperatures in Celsius for two cities over a week:

city1 = np.array([20, 22, 21, 19, 18, 20, 23])
city2 = np.array([30, 32, 31, 29, 28, 30, 33])

How would you use np.corrcoef() to find if the temperatures in the two cities move together, and what does the off-diagonal value represent?

hard

A. Use np.corrcoef(city1); off-diagonal is correlation with itself

B. Use np.corrcoef(city1 + city2); off-diagonal is sum of temperatures

C. Use np.corrcoef(city1 * city2); off-diagonal is product correlation

D. Use np.corrcoef(city1, city2); off-diagonal is correlation between city1 and city2 temperatures

Solution

1. Step 1: Apply np.corrcoef to two arrays

   Pass both city1 and city2 arrays as separate arguments: np.corrcoef(city1, city2).

2. Step 2: Interpret the off-diagonal value

   The off-diagonal value in the resulting matrix shows the correlation coefficient between the two cities' temperatures, indicating how they move together.

3. Final Answer:

   Use np.corrcoef(city1, city2); off-diagonal is correlation between city1 and city2 temperatures -> Option D

4. Quick Check:

   Off-diagonal = correlation between two arrays [OK]

Hint: Off-diagonal in np.corrcoef matrix shows correlation between inputs [OK]

Common Mistakes:

• Passing combined arrays instead of separate ones
• Misunderstanding off-diagonal meaning
• Using single array input expecting two-variable correlation