Matplotlib: Your First Charts

Numbers convince analysts; pictures convince everyone else. Matplotlib is Python's foundational plotting library — nearly every other charting tool (including seaborn, next lesson) is built on top of it. Here you'll learn the four workhorse chart types, how to label and style them, and how to arrange several plots in one figure.

The pyplot workflow

The standard import is import matplotlib.pyplot as plt. A minimal plot is three lines: prepare data, call a plot function, show the figure:

Python — runs in your browser

Run

import numpy as np import matplotlib.pyplot as plt x = np.linspace(0, 10, 100) # 100 smooth points y = np.sin(x) plt.plot(x, y) plt.show()

Every plt.* call between the plot function and plt.show() modifies the current figure — that's how you'll add titles, labels, and more lines. There is also an object-oriented style (fig, ax = plt.subplots() then ax.plot), which shines for multi-panel figures — we'll use it in the subplots section.

A fully dressed line chart

A chart without labels is a puzzle. Here's the checklist version — two lines, a title, axis labels, a legend, and a grid:

Python — runs in your browser

Run

import numpy as np import matplotlib.pyplot as plt x = np.linspace(0, 10, 100) plt.figure(figsize=(8, 4)) # width, height in inches plt.plot(x, np.sin(x), "r--", linewidth=2, label="sin(x)") plt.plot(x, np.cos(x), "b-.", linewidth=2, label="cos(x)") plt.xlim(0, 12) plt.ylim(-1.5, 1.5) plt.title("Trigonometric functions") plt.xlabel("x") plt.ylabel("y") plt.legend() plt.grid(alpha=0.3) plt.show()

The compact format strings pack color + marker + linestyle into a few characters:

• "r--" — red, dashed line
• "b-." — blue, dash-dot line
• "go" — green, circle markers (no line)
• "k^:" — black, triangle markers, dotted line

Prefer the explicit keywords (color="tab:red", linestyle="--", marker="o") in real code — they're easier to read six months later.

Scatter plots: relationships between two variables

Use a scatter plot when each point is one observation and you want to see how two quantities relate. plt.scatter can also encode a third variable via color and a fourth via size:

Python — runs in your browser

Run

import numpy as np import matplotlib.pyplot as plt rng = np.random.default_rng(42) hours = rng.uniform(0, 10, 60) # hours studied score = 40 + 5.5 * hours + rng.normal(0, 6, 60) # exam score attendance = rng.uniform(0.5, 1.0, 60) # share of classes attended plt.figure(figsize=(7, 4.5)) sc = plt.scatter(hours, score, c=attendance, s=attendance * 80, cmap="viridis", alpha=0.8) plt.colorbar(sc, label="attendance") plt.xlabel("hours studied") plt.ylabel("exam score") plt.title("Study time vs. exam score") plt.show()

The upward trend jumps out immediately — that's the point of a scatter plot.

Bar charts: comparing categories

Bars compare a numeric value across categories. Keep them honest: start the y-axis at zero and sort when order isn't meaningful:

Python — runs in your browser

Run

import matplotlib.pyplot as plt cities = ["Jakarta", "Surabaya", "Bandung", "Medan", "Makassar"] revenue = [420, 265, 310, 180, 150] plt.figure(figsize=(7, 4)) plt.bar(cities, revenue, color="tab:blue") plt.ylabel("revenue (million IDR)") plt.title("Revenue by city") plt.grid(axis="y", alpha=0.3) plt.show()

For long category names, plt.barh (horizontal bars) keeps labels readable.

Histograms: the shape of one variable

A histogram chops a numeric variable into bins and counts how many values land in each — it's the fastest way to see a distribution's shape, center, and outliers:

Python — runs in your browser

Run

import numpy as np import matplotlib.pyplot as plt rng = np.random.default_rng(0) uniform_data = rng.random(5000) # flat normal_data = rng.normal(0, 1, 5000) # bell curve plt.figure(figsize=(8, 4)) plt.hist(normal_data, bins=30, alpha=0.7, label="normal", color="tab:blue") plt.hist(uniform_data * 6 - 3, bins=30, alpha=0.5, label="uniform (rescaled)", color="tab:orange") plt.xlabel("value") plt.ylabel("count") plt.title("Two very different distributions") plt.legend() plt.show()

Try changing bins — too few hides structure, too many turns the plot into noise. Between 20 and 50 is a good starting range for a few thousand points.

Subplots: several charts, one figure

plt.subplots(rows, cols) returns a figure and an array of axes; each axis is its own little plotting surface with ax.plot, ax.set_title, and friends:

Python — runs in your browser

Run

import numpy as np import matplotlib.pyplot as plt rng = np.random.default_rng(1) x = np.linspace(0, 10, 100) fig, axes = plt.subplots(2, 2, figsize=(9, 6)) axes[0, 0].plot(x, np.sin(x), color="tab:blue") axes[0, 0].set_title("line") axes[0, 1].scatter(rng.random(50), rng.random(50), color="tab:orange", s=20) axes[0, 1].set_title("scatter") axes[1, 0].bar(["A", "B", "C"], [3, 7, 5], color="tab:green") axes[1, 0].set_title("bar") axes[1, 1].hist(rng.normal(size=800), bins=25, color="tab:red") axes[1, 1].set_title("histogram") fig.suptitle("Four chart types at a glance") fig.tight_layout() plt.show()

fig.tight_layout() fixes overlapping labels — make it a habit for any multi-panel figure.

Check your understanding

4 questions · free

1. Q1.Your data is average temperature per month over 3 years. Which chart type fits best?

   AHistogramBLine chartCPie chartDScatter with no ordering

2. Q2.What does the format string "r--" mean in plt.plot(x, y, "r--")?

   ARed dashed lineBRight-aligned labelsCTwo red markers per pointDReverse the x-axis twice

3. Q3.You call plt.hist(data, bins=3) on 10,000 values and the plot looks like three boring blocks. What should you do?

   AUse a pie chart insteadBIncrease the number of bins to reveal the distribution's shapeCSort the data firstDTake the logarithm of the counts

4. Q4.fig, axes = plt.subplots(2, 3) — how do you plot on the bottom-right panel?

   Aaxes[3, 2].plot(...)Baxes[1, 2].plot(...)Caxes[2, 3].plot(...)Dfig[1][2].plot(...)

Exercise: A two-panel sales dashboard

Build a 1×2 figure: on the left, a line chart of 12 months of simulated sales (an upward trend plus random noise, with circle markers); on the right, a histogram of 500 simulated order values (normal around 50, no negatives). Give every panel a title and axis labels, and make the whole figure figsize=(10, 4).

Matplotlib gives you full control; next lesson, seaborn gives you beautiful statistical charts with a fraction of the code.