Correlation matrix visualization in Matplotlib - Step-by-Step Execution

Concept Flow - Correlation matrix visualization

Start with DataFrame

↓

Calculate Correlation Matrix

↓

Create Heatmap Plot

↓

Add Color Legend

↓

Show Plot

↓

End

We start with data, calculate correlations, then visualize them as a heatmap with colors showing strength and direction.

Execution Sample

Matplotlib

import matplotlib.pyplot as plt
import pandas as pd
import numpy as np

data = pd.DataFrame(np.random.rand(5,5), columns=list('ABCDE'))
corr = data.corr()
plt.imshow(corr, cmap='coolwarm', vmin=-1, vmax=1)
plt.colorbar()
plt.show()

This code creates a random 5x5 data table, computes correlations, and shows them as a colored grid.

Execution Table

Step	Action	Variable/Output	Description
1	Create random data	data	5x5 table with random numbers between 0 and 1
2	Calculate correlation	corr	5x5 matrix showing correlation coefficients between columns
3	Plot correlation matrix	imshow plot	Heatmap with colors from blue (-1) to red (+1)
4	Add colorbar	colorbar	Shows color scale for correlation values
5	Display plot	plot window	Visual output of correlation heatmap
6	End	-	Visualization complete

💡 Plot shown and program ends

Variable Tracker

Variable	Start	After Step 1	After Step 2	After Step 3	Final
data	None	5x5 random numbers	5x5 random numbers	5x5 random numbers	5x5 random numbers
corr	None	None	5x5 correlation matrix	5x5 correlation matrix	5x5 correlation matrix
imshow plot	None	None	None	Heatmap created	Heatmap displayed

Key Moments - 3 Insights

Why do we use data.corr() before plotting?

Because the correlation matrix shows relationships between columns, and we need to calculate it first (see Step 2 in execution_table).

What does the color range from blue to red mean in the heatmap?

Blue means strong negative correlation (-1), red means strong positive correlation (+1), and colors in between show weaker correlations (Step 3).

Why do we add plt.colorbar() after imshow?

The colorbar explains what the colors represent in terms of correlation values, helping us read the heatmap (Step 4).

Visual Quiz - 3 Questions

Look at the execution_table at Step 2, what is stored in 'corr'?

A A color map for plotting

B A 5x5 matrix of correlation coefficients

C Random numbers between 0 and 1

D A plot window

💡 Hint

Check Step 2 in execution_table where 'corr' is described.

At which step is the heatmap plot actually created?

A Step 3

B Step 1

C Step 4

D Step 5

💡 Hint

Look at the 'Action' column in execution_table for when 'imshow plot' is created.

If we skip plt.colorbar(), what will be missing in the visualization?

A The heatmap colors

B The correlation matrix calculation

C The color scale legend explaining colors

D The data table

💡 Hint

Refer to Step 4 in execution_table about adding the colorbar.

Concept Snapshot

Correlation matrix visualization:
- Calculate correlation matrix with data.corr()
- Use plt.imshow() to create heatmap
- Set color map (e.g., 'coolwarm') and limits (-1 to 1)
- Add plt.colorbar() for color scale
- Call plt.show() to display plot

Full Transcript

We start with a data table and calculate the correlation matrix to find relationships between columns. Then we create a heatmap plot using matplotlib's imshow function, coloring cells by correlation values from -1 to 1. We add a colorbar to explain the colors. Finally, we show the plot window. This helps us visually understand how variables relate to each other.

Practice

1. What does a correlation matrix visualization primarily help you understand in data science?

easy

A. The total number of missing values in the data

B. The strength and direction of relationships between variables

C. The order of rows in a dataset

D. The exact values of each data point in the dataset

Solution

1. Step 1: Understand the purpose of a correlation matrix

   A correlation matrix shows how variables relate to each other, indicating strength and direction of relationships.

2. Step 2: Connect visualization to understanding relationships

   Visualizing the matrix with colors helps quickly spot strong positive or negative correlations between variables.

3. Final Answer:

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

4. Quick Check:

   Correlation matrix = relationships [OK]

Hint: Correlation matrix shows variable relationships, not raw data [OK]

Common Mistakes:

• Confusing correlation matrix with raw data values
• Thinking it shows missing data counts
• Assuming it orders dataset rows

2. Which of the following is the correct way to display a correlation matrix using matplotlib's imshow function?

easy

A. plt.imshow(corr_matrix, cmap='coolwarm', interpolation='none')

B. plt.imshow(corr_matrix, color='red')

C. plt.imshow(corr_matrix, style='heatmap')

D. plt.imshow(corr_matrix, axis='off')

Solution

1. Step 1: Recall correct imshow parameters for correlation matrix

   Use cmap to set color map and interpolation='none' to avoid smoothing colors.

2. Step 2: Evaluate options

   plt.imshow(corr_matrix, cmap='coolwarm', interpolation='none') correctly uses cmap='coolwarm' and interpolation='none'. Others use invalid or unsupported parameters.

3. Final Answer:

   plt.imshow(corr_matrix, cmap='coolwarm', interpolation='none') -> Option A

4. Quick Check:

   Correct imshow syntax = plt.imshow(corr_matrix, cmap='coolwarm', interpolation='none') [OK]

Hint: Use cmap and interpolation='none' with imshow for clear matrix [OK]

Common Mistakes:

• Using invalid parameters like 'color' or 'style' in imshow
• Forgetting interpolation='none' causing blurry colors
• Trying to hide axes with 'axis' parameter inside imshow

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

import matplotlib.pyplot as plt
import numpy as np

np.random.seed(0)
data = np.random.rand(4,4)
corr_matrix = np.corrcoef(data)
plt.imshow(corr_matrix, cmap='viridis', interpolation='none')
plt.colorbar()
plt.show()

medium

A. A line plot showing correlation coefficients over time

B. A scatter plot of the original data points

C. A heatmap showing correlation values between 4 variables with colors from viridis colormap

D. An error because np.corrcoef cannot handle 2D arrays

Solution

1. Step 1: Understand the code steps

   Data is a 4x4 random array, then np.corrcoef computes correlation matrix (4x4). imshow visualizes it with 'viridis' colors and a colorbar.

2. Step 2: Identify the output type

   The output is a heatmap showing correlation strengths between variables, colored by viridis scale.

3. Final Answer:

   A heatmap showing correlation values between 4 variables with colors from viridis colormap -> Option C

4. Quick Check:

   Correlation matrix + imshow = heatmap [OK]

Hint: Correlation matrix + imshow + colorbar = heatmap visualization [OK]

Common Mistakes:

• Confusing correlation matrix with scatter or line plots
• Thinking np.corrcoef fails on 2D arrays
• Ignoring the colorbar presence

4. You run this code but get a blurry correlation matrix image:

plt.imshow(corr_matrix, cmap='coolwarm')
plt.colorbar()
plt.show()

What is the likely cause and fix?

medium

A. The colormap 'coolwarm' is invalid; use 'viridis' instead

B. The correlation matrix must be normalized before plotting

C. You forgot to call plt.show(), so image is blurry

D. Missing interpolation='none' causes smoothing; add interpolation='none' to fix

Solution

1. Step 1: Identify cause of blurry image

   By default, imshow applies interpolation which smooths colors, causing blur.

2. Step 2: Fix by disabling interpolation

   Adding interpolation='none' disables smoothing, making each cell color sharp and clear.

3. Final Answer:

   Missing interpolation='none' causes smoothing; add interpolation='none' to fix -> Option D

4. Quick Check:

   Interpolation='none' stops blur [OK]

Hint: Add interpolation='none' to imshow to avoid blurry colors [OK]

Common Mistakes:

• Changing colormap instead of fixing interpolation
• Forgetting plt.show() does not affect blur
• Normalizing matrix is unnecessary for visualization

5. You have a dataset with 5 variables and want to visualize their correlation matrix with variable names on axes. Which code snippet correctly adds labels and a title to the heatmap?

hard

A. plt.imshow(corr_matrix, cmap='coolwarm', interpolation='none')
plt.xticks(ticks=range(5), labels=var_names)
plt.yticks(ticks=range(5), labels=var_names)
plt.title('Correlation Matrix')
plt.colorbar()
plt.show()

B. plt.imshow(corr_matrix, cmap='coolwarm')
plt.xlabel(var_names)
plt.ylabel(var_names)
plt.title('Correlation Matrix')
plt.colorbar()
plt.show()

C. plt.imshow(corr_matrix, cmap='coolwarm', interpolation='none')
plt.axis('off')
plt.title('Correlation Matrix')
plt.colorbar()
plt.show()

D. plt.imshow(corr_matrix, cmap='coolwarm', interpolation='none')
plt.xticks(var_names)
plt.yticks(var_names)
plt.colorbar()
plt.show()

Solution

1. Step 1: Understand how to label axes in matplotlib heatmap

   Use plt.xticks and plt.yticks with ticks=range(n) and labels=var_names to set variable names on axes.

2. Step 2: Check options for correct labeling and title

   plt.imshow(corr_matrix, cmap='coolwarm', interpolation='none') plt.xticks(ticks=range(5), labels=var_names) plt.yticks(ticks=range(5), labels=var_names) plt.title('Correlation Matrix') plt.colorbar() plt.show() correctly sets ticks and labels, adds title and colorbar. Others misuse xlabel, ylabel, or omit ticks.

3. Final Answer:

   plt.imshow(corr_matrix, cmap='coolwarm', interpolation='none') plt.xticks(ticks=range(5), labels=var_names) plt.yticks(ticks=range(5), labels=var_names) plt.title('Correlation Matrix') plt.colorbar() plt.show() -> Option A

4. Quick Check:

   Use xticks/yticks with ticks and labels [OK]

Hint: Use plt.xticks and plt.yticks with ticks and labels for axis names [OK]

Common Mistakes:

• Using plt.xlabel/plt.ylabel for multiple tick labels
• Forgetting to specify ticks in xticks/yticks
• Turning axis off removes labels