A manager will look at your chart for 5 seconds. In that time it must answer one question clearly – “Are sales rising?”, “Which city is best?” – without explanation. Matplotlib gives you full control to make that happen.
Which chart for which question?
📈
LineChange over time – monthly sales, temperature.
📊
BarCompare categories – revenue by city.
🔵
ScatterRelationship between two numbers – area vs price.
📶
HistogramDistribution of one number – customer ages.
🥧
Pie / donutShare of a whole, max 5 parts.
Pick the chart from the question
Figure and Axes
A Figure is the whole image; an Axes is one chart inside it. Always use fig, ax = plt.subplots() and call methods on ax – it scales cleanly to many charts.
Grey for the past, a strong colour for “now”, and a title that states the conclusion.
Bar charts
Python
import pandas as pd
import matplotlib.pyplot as plt
s = pd.read_csv("sales_data.csv")
s["revenue"] = s["quantity"] * s["unit_price"] * (1 - s["discount"])
city = (s.groupby("city")["revenue"].sum() / 1e5).sort_values()
colors = ["#4f46e5" if v == city.max() else "#c7d2fe" for v in city]
fig, ax = plt.subplots(figsize=(8, 4.5))
bars = ax.barh(city.index, city.values, color=colors)
ax.bar_label(bars, fmt="%.1f", padding=3, fontsize=8)
ax.set_title("Mumbai leads revenue (₹ lakh, 2024)"); ax.set_xlabel("₹ lakh")
ax.spines[["top", "right"]].set_visible(False)
plt.show()
Sorted bars + one highlight colour = the answer in one glance.
Python
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
s = pd.read_csv("sales_data.csv")
pv = pd.crosstab(s["category"], s["customer_gender"])
x = np.arange(len(pv)); w = 0.38
fig, ax = plt.subplots(figsize=(9, 4))
ax.bar(x - w / 2, pv["Female"], w, label="Female", color="#ec4899")
ax.bar(x + w / 2, pv["Male"], w, label="Male", color="#3b82f6")
ax.set_xticks(x, pv.index, rotation=10); ax.set_ylabel("Orders"); ax.legend()
ax.set_title("Orders by category and gender")
plt.show()
Grouped bars compare two groups inside each category.
Scatter and histogram
Python
import pandas as pd
import matplotlib.pyplot as plt
h = pd.read_csv("house_prices.csv")
fig, ax = plt.subplots(1, 2, figsize=(11, 4))
for city, color in [("Bengaluru", "#4f46e5"), ("Lucknow", "#f59e0b")]:
d = h[h["city"] == city]
ax[0].scatter(d["area_sqft"], d["price_lakh"], s=14, alpha=0.6, color=color, label=city)
ax[0].set_xlabel("Area (sqft)"); ax[0].set_ylabel("₹ lakh"); ax[0].legend(); ax[0].set_title("Same area, different city price")
s = pd.read_csv("sales_data.csv")
ax[1].hist(s["customer_age"].dropna(), bins=20, color="#0891b2", edgecolor="white")
ax[1].axvline(s["customer_age"].median(), color="#dc2626", ls="--", label="median")
ax[1].set_title("Customer age distribution"); ax[1].set_xlabel("Age"); ax[1].legend()
plt.tight_layout(); plt.show()
Colour by group in a scatter; mark the median on a histogram.
Subplots: a mini dashboard
Python
import pandas as pd
import matplotlib.pyplot as plt
s = pd.read_csv("sales_data.csv", parse_dates=["order_date"])
s["revenue"] = s["quantity"] * s["unit_price"] * (1 - s["discount"])
fig, ax = plt.subplots(2, 2, figsize=(11, 7))
m = s.groupby(s["order_date"].dt.month)["revenue"].sum() / 1e5
ax[0, 0].plot(m.index, m.values, marker="o", color="#4f46e5"); ax[0, 0].set_title("Revenue by month (₹ lakh)")
c = s.groupby("category")["revenue"].sum().sort_values()
ax[0, 1].barh(c.index, c / 1e5, color="#0891b2"); ax[0, 1].set_title("Revenue by category")
p = s["payment_mode"].value_counts()
ax[1, 0].pie(p, labels=p.index, autopct="%1.0f%%", wedgeprops=dict(width=0.45)); ax[1, 0].set_title("Payment mode")
ax[1, 1].hist(s["rating"].dropna(), bins=[1.75, 2.25, 2.75, 3.25, 3.75, 4.25, 4.75, 5.25], color="#16a34a", edgecolor="white"); ax[1, 1].set_title("Ratings")
fig.suptitle("BazaarKart 2024 at a glance", fontsize=15, fontweight="bold")
plt.tight_layout(); plt.savefig("dashboard.png", dpi=150); plt.show()
plt.savefig("dashboard.png", dpi=150) saves it for a report or slide.
7 rules for professional charts
The title states the insight, not just the topic.
Label axes with units (₹ lakh, %, days).
Sort bars by value.
Use one highlight colour; grey for the rest.
Start bar axes at zero.
Avoid 3D and pies with more than 5 slices.
Remove clutter: top/right borders, heavy grids.
🏅 Job skills unlocked today
✔ Use fig, ax = plt.subplots()✔ Choose and build line, bar, scatter, hist, pie✔ Annotate and highlight✔ Build and save multi-panel figures
Practice tasks
🏋️ Practice – do it yourself
Task 1: Plot the monthly average temperature of Delhi for 2024 as a line, and shade the summer months (Apr–Jun) with ax.axvspan.Show solution
Python
import pandas as pd
import matplotlib.pyplot as plt
w = pd.read_csv("delhi_weather.csv", parse_dates=["date"])
t = w[w["date"].dt.year == 2024].groupby(w["date"].dt.month)["avg_temp_c"].mean()
fig, ax = plt.subplots(figsize=(8, 3.5))
ax.plot(t.index, t.values, marker="o", color="#dc2626")
ax.axvspan(4, 6, color="#fde68a", alpha=0.5, label="Summer")
ax.set_title("Delhi 2024: hottest in May–June"); ax.set_xlabel("Month"); ax.set_ylabel("°C"); ax.legend()
plt.show()
Output of the program
Task 2: Make a horizontal bar chart of the churn rate by contract type from the telecom data, highlighting the worst one in red.Show solution
Python
import pandas as pd
import matplotlib.pyplot as plt
t = pd.read_csv("telecom_churn.csv")
r = (t.groupby("contract")["churn"].apply(lambda c: (c == "Yes").mean()) * 100).sort_values()
fig, ax = plt.subplots(figsize=(7, 3))
ax.barh(r.index, r.values, color=["#dc2626" if v == r.max() else "#cbd5e1" for v in r])
ax.set_title("Monthly contracts churn the most (%)"); ax.set_xlabel("Churn %")
plt.show()
Output of the program
Task 3: Recreate the 2×2 dashboard for students_performance.csv: histogram of average, bar of mean score by school type, scatter of study hours vs average, pie of internet access.
Quick quiz
🧠 Quick Quiz – test yourself
1. Best chart to show revenue over 12 months?
Time on the x-axis → line.
2. fig, ax = plt.subplots(2, 3) creates…
ax is a 2×3 array: ax[0, 0] … ax[1, 2].
3. Which adds value labels on bars?
Pass the container returned by ax.bar/barh.
4. A good chart title is…
State the insight.
5. To save a chart call savefig…
show() may clear the figure, so save first.
Interview corner
💼 Interview Corner
How do you decide which chart to use?
Start from the question: trend over time → line; compare categories → sorted bar; relationship → scatter; distribution → histogram/box; part of whole with few parts → stacked bar or donut. Then remove clutter and highlight the key point.
Why might a bar chart that does not start at zero be misleading?
Bar length encodes the value. Cutting the axis exaggerates small differences – a 2% gap can look like 50%.
