06 · Plotting & Visualization Basics

MATLAB's plotting is one of its strongest, most heavily used features — built-in, fast, and designed so a single plot() call produces a usable figure with almost no configuration.

The basic plot()

x = 0:0.1:2*pi;
y = sin(x);
plot(x, y)

This opens a Figure window showing a smooth sine curve. x and y must be the same length — plot() connects them point by point in order. Nothing is printed to the Command Window; the output is entirely visual.

Labels, title, and grid — always add these

A plot without axis labels is close to useless for anyone but the person who made it five minutes ago:

x = 0:0.1:2*pi;
y = sin(x);
plot(x, y)
xlabel('x (radians)')
ylabel('sin(x)')
title('Sine Wave')
grid on

grid on overlays light gridlines, which make it much easier to read values off the curve. These four lines — plot, xlabel, ylabel, title (plus usually grid on) — are the baseline for almost every MATLAB plot you'll make.

Multiple lines on one plot

x = 0:0.1:2*pi;
plot(x, sin(x), x, cos(x))
legend('sin(x)', 'cos(x)')
xlabel('x')
ylabel('y')
title('Sine and Cosine')

Passing multiple x, y pairs to one plot() call draws them all on the same axes, each with a different default color, and legend() labels them in the order they were plotted.

hold on is the alternative approach — useful when the curves come from separate plot() calls (e.g. inside a loop) rather than one call:

plot(x, sin(x))
hold on               % keep the current plot; don't erase it for the next one
plot(x, cos(x))
hold off               % (optional) return to default overwrite behavior
legend('sin(x)', 'cos(x)')

Forgetting hold on

Without hold on, each new plot() call erases the previous figure content and starts fresh — a very common beginner mistake when trying to build up a chart across multiple calls or a loop, where only the last curve ends up visible.

Line styles, colors, and markers

A short format string as the third argument to plot() controls appearance:

plot(x, sin(x), 'r--')      % red dashed line
plot(x, sin(x), 'g-o')      % green solid line with circle markers
plot(x, sin(x), 'b:', 'LineWidth', 2)   % blue dotted, thicker line (name-value pair)

Code	Meaning	Code	Meaning
r	red	-	solid line
g	green	--	dashed line
b	blue	:	dotted line
k	black	-.	dash-dot line
o	circle marker	x	x marker

Format-string characters can combine freely ('r--o' = red, dashed, with circle markers); properties that don't have a short code, like LineWidth or MarkerSize, are set as trailing 'Name', value pairs.

Scatter plots

scatter() is the right choice for unconnected, discrete data points — where drawing a line between them (as plot does) would misleadingly imply an ordering or trend:

hours_studied = [1, 2, 3, 4, 5, 6, 7, 8];
exam_score    = [52, 58, 63, 70, 74, 82, 88, 91];
scatter(hours_studied, exam_score, 'filled')
xlabel('Hours Studied')
ylabel('Exam Score')
title('Study Time vs. Score')
grid on

Bar charts and histograms

categories = {'Q1', 'Q2', 'Q3', 'Q4'};
revenue    = [120, 150, 90, 200];
bar(revenue)
set(gca, 'XTickLabel', categories)   % label the bars with category names
ylabel('Revenue ($k)')
title('Quarterly Revenue')

bar() is for a small number of discrete categories with known values. histogram() is different: it takes a set of raw, possibly-many observations and automatically bins them, showing you the underlying distribution:

data = randn(1, 1000);     % 1000 random values from a standard normal distribution
histogram(data)
xlabel('Value')
ylabel('Count')
title('Distribution of Random Data')

Multiple plots in one figure: subplot

x = 0:0.1:2*pi;

subplot(2, 1, 1)      % 2 rows, 1 column, this is plot #1
plot(x, sin(x))
title('Sine')

subplot(2, 1, 2)      % 2 rows, 1 column, this is plot #2
plot(x, cos(x))
title('Cosine')

subplot(rows, cols, index) divides the current figure into a grid and selects one cell to draw into next; index counts left-to-right, top-to-bottom, same convention as reading English text.

Saving a figure to a file

plot(x, sin(x))
xlabel('x'); ylabel('sin(x)'); title('Sine Wave')
saveas(gcf, 'sine_wave.png')     % gcf = "get current figure"

saveas infers the file format from the extension (.png, .jpg, .pdf, .fig for MATLAB's own editable format, etc.). exportgraphics(gcf, 'sine_wave.png', 'Resolution', 300) is the more modern, higher-quality alternative when you need print-resolution output.

New vs. reused figures

figure               % opens a brand-new figure window
plot(x, sin(x))

figure               % opens ANOTHER new window, doesn't touch the first
plot(x, cos(x))

Without an explicit figure call, MATLAB reuses the current figure window (and, per the hold on/hold off rule above, overwrites its contents by default) — call figure explicitly whenever you want a fresh, separate window.

Plotting cheat sheet

Task	Function
Line plot	plot(x, y)
Scatter (unconnected points)	scatter(x, y)
Bar chart (categories)	bar(values)
Histogram (distribution of raw data)	histogram(data)
Axis labels / title	xlabel(), ylabel(), title()
Multiple series, one legend	legend('name1', 'name2')
Keep adding to same axes	hold on / hold off
Gridlines	grid on
New figure window	figure
Multiple plots, one figure	subplot(rows, cols, index)
Save to file	saveas(gcf, 'name.png')

How It Actually Works

Every plot() call builds an in-memory graphics object tree, not an image: calling plot(x,y) creates (or reuses) a Figure object, which contains an Axes object, which contains a Line object holding your x/y data plus style properties (Color, LineWidth, Marker, …). Nothing is rasterized to pixels until the rendering pipeline actually needs to paint the screen or export a file — this is why you can grab a handle (h = plot(x,y)) and change h.Color after the plot has already "appeared": you're mutating a live object in the graphics tree, and MATLAB's renderer redraws from that tree, not from a frozen bitmap.

hold on doesn't change how plotting math works — it flips a property (NextPlot) on the current Axes object from 'replace' to 'add', telling subsequent plotting commands to append new graphics-object children to the existing axes instead of deleting the old ones first. Axis limits (xlim, ylim) are computed automatically by default via a "tight to data, then padded" heuristic that re-runs every time a child object is added or removed, unless you pin them manually — which is why adding a second plot() call to the same axes can silently rescale your first curve's apparent shape as the axes auto-fit both datasets.

Rendering itself goes through one of two backends: the OpenGL-accelerated renderer (default for anything with transparency, lighting, or many points) or the software painters renderer (default for simple 2D vector content, and what's used when exporting to vector formats like PDF/EPS so lines stay crisp rather than being rasterized).

Note: describes MATLAB's documented graphics-object model; there is no MATLAB installation in this environment to render an actual figure.

🔀 See this in another language

• SQL — Joins Basics
• TypeScript — Arrays & Objects, Typed
• Ruby — Strings & String Methods

Exercise

Create x = linspace(0, 10, 100) and plot y1 = x.^2 and y2 = 10*x on the same axes using hold on, with a legend distinguishing the two curves, axis labels, a title, and grid on. Then use subplot to create a 1×2 figure: the left panel showing your line plot from above, and the right panel showing a histogram of 500 values from randn(1, 500). Save the full figure to exercise_plot.png using saveas.