Scatter plots in MATLAB - Deep Dive

Overview - Scatter plots

What is it?

A scatter plot is a simple graph that shows how two sets of numbers relate to each other. Each point on the plot represents one pair of values, with its position showing the values on the horizontal and vertical axes. This helps us see patterns, trends, or groups in data. It is often used to explore relationships between two variables.

Why it matters

Scatter plots help us quickly understand if two things are connected or not, like height and weight or temperature and ice cream sales. Without scatter plots, spotting these connections would be slow and confusing, especially with lots of data. They make data visual and easy to grasp, which is crucial for making good decisions based on data.

Where it fits

Before learning scatter plots, you should know basic plotting and how to handle simple data arrays in MATLAB. After mastering scatter plots, you can explore more complex visualizations like line plots, histograms, and 3D plots, or learn statistical methods to measure relationships seen in scatter plots.

Mental Model

Core Idea

A scatter plot places each pair of numbers as a dot on a flat surface, letting you see how two things change together.

Think of it like...

Imagine throwing a handful of small balls onto a flat table where the table's length and width represent two different measurements. Where each ball lands shows the combination of those two measurements for one item.

  Y-axis
    ↑
    │       ●       ●
    │   ●       ●
    │       ●
    │  ●
    │
    └────────────────→ X-axis
   Each ● is a data point showing two values together

Build-Up - 7 Steps

1

Foundation

Understanding basic data pairs

Concept: Learn what pairs of numbers represent and how they relate to two variables.

Data for a scatter plot comes as pairs, like (x, y). For example, x could be hours studied, and y could be test scores. Each pair shows one student's hours and score.

Result

You understand that each pair is one point to plot, linking two measurements.

Understanding data as pairs is the foundation for plotting relationships between two variables.

2

Foundation

Creating a simple scatter plot in MATLAB

Concept: Learn the basic MATLAB command to plot pairs of data points.

Use the scatter function: scatter(x, y) where x and y are vectors of the same length. For example: x = [1 2 3 4 5]; y = [2 4 1 3 5]; scatter(x, y); title('Simple Scatter Plot'); xlabel('X values'); ylabel('Y values');

Result

A window opens showing dots at the positions given by x and y values.

Knowing the scatter function lets you quickly visualize data pairs and see their distribution.

3

Intermediate

Customizing scatter plot appearance

🤔Before reading on: Do you think you can change the color and size of points in scatter plots? Commit to yes or no.

Concept: Learn how to change point colors, sizes, and markers to make plots clearer or more informative.

You can add extra arguments to scatter, like: scatter(x, y, 100, 'r', 'filled'); This makes points bigger (size 100), red ('r'), and filled circles. You can also use different markers like 'o', '+', '*'.

Result

The plot shows bigger red filled circles instead of default small blue dots.

Customizing points helps highlight important data or groups, improving communication of insights.

4

Intermediate

Adding labels and grid for clarity

🤔Before reading on: Do you think adding labels and grid lines helps interpret scatter plots better? Commit to yes or no.

Concept: Learn to add axis labels, titles, and grid lines to make plots easier to understand.

Use xlabel('X label'), ylabel('Y label'), title('Plot title'), and grid on to add these features: scatter(x, y); xlabel('Hours Studied'); ylabel('Test Score'); title('Study Hours vs Test Scores'); grid on;

Result

The plot now has clear axis names, a title, and grid lines to help read values.

Labels and grids guide the viewer’s eye and make the plot’s story clearer.

5

Intermediate

Plotting multiple groups with colors

🤔Before reading on: Can you plot different groups in one scatter plot using colors? Commit to yes or no.

Concept: Learn to show different categories by assigning colors to points based on group membership.

Suppose you have two groups: x = [1 2 3 4 5 6 7 8]; y = [2 4 1 3 5 7 6 8]; groups = [1 1 1 1 2 2 2 2]; colors = ['r', 'b']; figure; hold on; for g = 1:2 scatter(x(groups==g), y(groups==g), 100, colors(g), 'filled'); end hold off; title('Scatter Plot with Groups');

Result

The plot shows red points for group 1 and blue points for group 2.

Using colors for groups reveals patterns or differences between categories in data.

6

Advanced

Adding trend lines to scatter plots

🤔Before reading on: Do you think scatter plots can show trends with lines? Commit to yes or no.

Concept: Learn to add a line that summarizes the relationship between x and y, like a best-fit line.

Calculate a linear fit: p = polyfit(x, y, 1); yfit = polyval(p, x); scatter(x, y, 100, 'filled'); hold on; plot(x, yfit, '-k', 'LineWidth', 2); hold off; title('Scatter Plot with Trend Line');

Result

The plot shows points and a black line that best fits the data trend.

Trend lines help quantify and visualize the overall relationship between variables.

7

Expert

Handling large datasets with scatter plots

🤔Before reading on: Do you think plotting millions of points directly is effective? Commit to yes or no.

Concept: Learn techniques to visualize very large data without clutter, like transparency or hexbin plots.

For large data, use transparency: scatter(x, y, 10, 'filled', 'MarkerFaceAlpha', 0.1); This makes points partly see-through, showing dense areas darker. MATLAB also supports hexbin plots via third-party tools for density visualization.

Result

The plot reveals dense clusters without overwhelming with overlapping points.

Managing visual clutter in big data is key to extracting meaningful patterns from scatter plots.

Under the Hood

MATLAB scatter plots work by mapping each pair of data values to coordinates on a 2D plane. Internally, the scatter function creates graphical objects called markers at these coordinates. These markers have properties like size, color, and shape, which MATLAB renders on the figure window. The rendering engine efficiently handles drawing and refreshing these points, even when customized.

Why designed this way?

Scatter plots were designed to visually represent relationships between two variables simply and intuitively. MATLAB's scatter function uses vectorized inputs to allow fast plotting of many points. The design balances ease of use with flexibility, enabling users to customize appearance without complex code. Alternatives like line plots or bar charts do not show individual data pairs as clearly.

Data pairs (x,y) → scatter function → graphical markers → figure window

┌─────────────┐     ┌───────────────┐     ┌───────────────┐     ┌───────────────┐
│ Numeric x,y │ --> │ scatter call  │ --> │ Marker objects│ --> │ Rendered plot │
└─────────────┘     └───────────────┘     └───────────────┘     └───────────────┘

Myth Busters - 4 Common Misconceptions

Quick: Does a scatter plot always show a cause-effect relationship? Commit to yes or no.

Common Belief: Scatter plots prove that one variable causes changes in the other.

Reality: Scatter plots only show that two variables move together, not that one causes the other.

Why it matters: Mistaking correlation for causation can lead to wrong decisions or false conclusions.

Quick: Can you use scatter plots for data with only one variable? Commit to yes or no.

Common Belief: Scatter plots work with any data, even if there is only one variable.

Reality: Scatter plots require two variables to plot pairs; one variable alone cannot form a scatter plot.

Why it matters: Trying to plot single-variable data as scatter plots wastes time and causes confusion.

Quick: Do larger scatter plot points always mean higher values? Commit to yes or no.

Common Belief: Bigger points on a scatter plot mean bigger data values.

Reality: Point size is a visual choice and can represent a third variable or be uniform; size does not automatically reflect data magnitude.

Why it matters: Misreading point size can lead to misunderstanding the data story.

Quick: Does overlapping points in scatter plots mean duplicate data? Commit to yes or no.

Common Belief: If points overlap exactly, the data must be duplicated.

Reality: Points can overlap because different data pairs have similar or identical values, not necessarily duplicates.

Why it matters: Assuming duplicates without checking can cause incorrect data cleaning or analysis.

Expert Zone

1

Scatter plots can be combined with marginal histograms to show distributions along each axis, revealing more about data spread.

2

Using transparency (alpha) in scatter plots helps visualize dense data regions without losing individual points.

3

Color mapping in scatter plots can encode continuous variables, adding a third dimension of information effectively.

When NOT to use

Scatter plots are not suitable when you have categorical data without numeric pairs or when you want to show trends over time clearly; line plots or bar charts are better alternatives.

Production Patterns

In real-world data science, scatter plots are often used in exploratory data analysis to detect outliers, clusters, or correlations before applying statistical models or machine learning.

Connections

Correlation coefficient

Scatter plots visually show relationships that correlation coefficients quantify numerically.

Understanding scatter plots helps interpret what correlation numbers mean in terms of actual data distribution.

Heatmaps

Heatmaps and scatter plots both visualize data density but heatmaps use color intensity over areas instead of points.

Knowing scatter plots clarifies how heatmaps summarize dense data regions differently.

Ecology - Species distribution

Scatter plots are used in ecology to map species locations by coordinates, showing spatial patterns.

Recognizing scatter plots in ecology reveals how data science tools apply to real-world environmental studies.

Common Pitfalls

#1 Plotting x and y vectors of different lengths.

Wrong approach:

x = [1 2 3 4];
y = [5 6 7];
scatter(x, y);

Correct approach:

x = [1 2 3 4];
y = [5 6 7 8];
scatter(x, y);

Root cause:

MATLAB requires x and y to have the same number of elements to pair points correctly.

#2 Using plot instead of scatter for discrete points.

Wrong approach:

x = [1 2 3];
y = [4 5 6];
plot(x, y, 'o');

Correct approach:

x = [1 2 3];
y = [4 5 6];
scatter(x, y);

Root cause:

plot connects points with lines by default, which can mislead interpretation of discrete data.

#3 Ignoring axis labels and titles.

Wrong approach:

scatter(x, y);

Correct approach:

scatter(x, y);
xlabel('Hours');
ylabel('Scores');
title('Study Hours vs Scores');

Root cause:

Without labels, viewers cannot understand what the axes or points represent.

Key Takeaways

Scatter plots show pairs of data points on two axes to reveal relationships visually.

Customizing point size, color, and labels improves clarity and insight from scatter plots.

Scatter plots do not prove cause and effect; they only show how variables move together.

Handling large datasets requires techniques like transparency to avoid clutter in scatter plots.

Scatter plots are a foundational tool in data science for exploring and communicating data patterns.