Heatmaps for correlation in Data Analysis Python - Time & Space Complexity

Time Complexity: Heatmaps for correlation

O(n²)

Understanding Time Complexity

We want to understand how the time to create a heatmap for correlation grows as the data size increases.

Specifically, how does the number of operations change when we calculate and display correlations for more variables?

Scenario Under Consideration

Analyze the time complexity of the following code snippet.

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

data = pd.DataFrame({f'var{i}': range(1000) for i in range(10)})

corr = data.corr()
sns.heatmap(corr)
plt.show()

This code creates a correlation heatmap for 10 variables, each with 1000 data points.

Identify Repeating Operations

Identify the loops, recursion, array traversals that repeat.

• Primary operation: Calculating pairwise correlations between variables.
• How many times: For each pair of variables, correlation is computed once, so roughly n x n times for n variables.

How Execution Grows With Input

As the number of variables increases, the number of correlation calculations grows quickly.

Input Size (n variables)	Approx. Operations (correlations)
10	100
100	10,000
1000	1,000,000

Pattern observation: The operations grow roughly with the square of the number of variables.

Final Time Complexity

Time Complexity: O(n²)

This means if you double the number of variables, the work to compute correlations roughly quadruples.

Common Mistake

[X] Wrong: "Calculating correlations grows linearly with the number of variables."

[OK] Correct: Each variable pairs with every other variable, so the number of pairs grows much faster than the number of variables.

Interview Connect

Understanding how correlation heatmaps scale helps you explain performance when working with many variables in real data projects.

Self-Check

"What if we only calculate correlations for a subset of variable pairs? How would the time complexity change?"

Practice

1. What does a heatmap of correlation typically show in data analysis?

easy

A. The total sum of all numerical columns

B. The frequency of each value in a dataset

C. The order of rows in a DataFrame

D. The strength and direction of relationships between variables

Solution

1. Step 1: Understand correlation meaning

   Correlation measures how two variables move together, either positively or negatively.

2. Step 2: Relate heatmap to correlation

   A heatmap uses colors to show these correlation strengths and directions between variables.

3. Final Answer:

   The strength and direction of relationships between variables -> Option D

4. Quick Check:

   Heatmap = Correlation strength/direction [OK]

Hint: Heatmaps color-code relationships between variables [OK]

Common Mistakes:

• Confusing correlation with frequency counts
• Thinking heatmaps show raw data values
• Assuming heatmaps show data order

2. Which of the following is the correct way to create a heatmap of correlations using seaborn in Python?

easy

A. sns.heatmap(data.corr())

B. sns.heatmap(data.corr)

C. sns.heatmap(corr(data))

D. sns.heatmap(data.correlation())

Solution

1. Step 1: Recall pandas correlation method

   In pandas, correlation is calculated by calling data.corr() with parentheses.

2. Step 2: Use seaborn heatmap correctly

   Seaborn's heatmap() accepts a matrix like the correlation matrix from data.corr().

3. Final Answer:

   sns.heatmap(data.corr()) -> Option A

4. Quick Check:

   Use parentheses for corr() method [OK]

Hint: Use data.corr() with parentheses inside sns.heatmap() [OK]

Common Mistakes:

• Omitting parentheses after corr
• Using non-existent corr() functions
• Passing raw data instead of correlation matrix

3. Given the following code, what will be the output of corr_matrix?

import pandas as pd
import numpy as np
data = pd.DataFrame({
  'A': [1, 2, 3, 4],
  'B': [4, 3, 2, 1],
  'C': [1, 3, 5, 7]
})
corr_matrix = data.corr()

medium

A. An error because data has different ranges

B. A matrix showing negative correlation between A and B, positive between A and C

C. A matrix with only positive correlations

D. A matrix showing all zeros because data is small

Solution

1. Step 1: Analyze data columns

   Column A increases, B decreases (inverse), C increases faster than A.

2. Step 2: Understand correlation signs

   A and B have negative correlation, A and C have positive correlation, B and C negative correlation.

3. Final Answer:

   A matrix showing negative correlation between A and B, positive between A and C -> Option B

4. Quick Check:

   Opposite trends = negative correlation [OK]

Hint: Opposite value trends mean negative correlation [OK]

Common Mistakes:

• Assuming small data means zero correlation
• Ignoring sign of correlation
• Expecting errors from different value ranges

4. You wrote this code to plot a heatmap but get an error:

import seaborn as sns
import pandas as pd
data = pd.DataFrame({'X':[1,2,3],'Y':[4,5,6]})
sns.heatmap(data.corr)

What is the error and how to fix it?

medium

A. DataFrame columns must be numeric for heatmap

B. Seaborn heatmap does not accept correlation matrices

C. Missing parentheses after corr; fix by using data.corr()

D. Missing import matplotlib.pyplot as plt

Solution

1. Step 1: Identify the error cause

   Calling data.corr without parentheses passes a method, not the correlation matrix.

2. Step 2: Correct the method call

   Add parentheses to call the method: data.corr() to get the matrix.

3. Final Answer:

   Missing parentheses after corr; fix by using data.corr() -> Option C

4. Quick Check:

   Method call needs parentheses [OK]

Hint: Always use parentheses to call methods like corr() [OK]

Common Mistakes:

• Passing method instead of result
• Assuming heatmap can't use correlation matrix
• Forgetting to import seaborn

5. You have a DataFrame with columns 'Age', 'Income', 'SpendingScore', and 'Savings'. You want to create a heatmap showing only correlations above 0.5 to focus on strong relationships. Which code snippet correctly creates this filtered heatmap?

hard

A. corr = data.corr(); sns.heatmap(corr.where(corr > 0.5))

B. corr = data.corr(); sns.heatmap(corr > 0.5)

C. corr = data.corr(); sns.heatmap(corr.filter(lambda x: x > 0.5))

D. corr = data.corr(); sns.heatmap(corr.dropna(thresh=0.5))

Solution

1. Step 1: Calculate correlation matrix

   Use data.corr() to get all correlations.

2. Step 2: Filter correlations above 0.5

   Use corr.where(corr > 0.5) to keep values above 0.5 and set others to NaN.

3. Step 3: Plot heatmap with filtered matrix

   Pass the filtered matrix to sns.heatmap() to visualize only strong correlations.

4. Final Answer:

   corr = data.corr(); sns.heatmap(corr.where(corr > 0.5)) -> Option A

5. Quick Check:

   Use where() to filter matrix values [OK]

Hint: Use corr.where(condition) to filter heatmap values [OK]

Common Mistakes:

• Using filter() which works on columns, not matrix values
• Trying to index matrix with boolean mask directly
• Dropping rows instead of masking values