Plotting with Pandas

The course notebooks end almost every analysis with a chart: ipl3['year'].value_counts().sort_index().plot(kind='bar'), df['Brand Name'].value_counts().head().plot(kind='pie'). A table of 10 numbers is hard to read; a bar chart of the same numbers is understood in one second. Pandas wraps Matplotlib so that any Series or DataFrame can draw itself with .plot().

Note: charts can't be shown in this site's text playground, so the examples on this page are marked not runnable. Run them in Jupyter Notebook / Google Colab, where the chart appears below the cell (add plt.show() in a plain .py script).

Definition

DataFrame.plot() / Series.plot() is pandas' built-in plotting method. It draws a Matplotlib chart from the data — the index becomes the x-axis and each numeric column becomes a series (line, bar, …) — and returns the Matplotlib Axes object so you can customise it further.

• Built on Matplotlib — you need matplotlib installed; pandas just calls it for you.
• Index → x-axis — this is why value_counts() and groupby(...).sum() results plot so nicely: their index is the category.
• Each column → one series — a DataFrame with 3 numeric columns draws 3 lines (or 3 bars per group).
• Returns Axes — you can call ax.set_title(...), ax.set_ylabel(...) on it.

Syntax

import matplotlib.pyplot as plt

df.plot(kind="line", x=None, y=None, title=None, figsize=None,
        legend=True, grid=False, color=None, rot=None, subplots=False)
df.plot.bar(...)        # same as df.plot(kind="bar", ...)
plt.show()              # needed in scripts; optional in Jupyter

Chart kinds — what each shows and when to use it

Example — bar chart of wins per team

import pandas as pd
import matplotlib.pyplot as plt

wins = pd.Series({"CSK": 5, "MI": 5, "KKR": 3, "RCB": 2, "SRH": 2, "DC": 1})
ax = wins.plot(kind="bar", title="Wins per team", color="steelblue", rot=0)
ax.set_xlabel("Team")
ax.set_ylabel("Wins")
plt.show()

What you see: six vertical bars, team codes along the x-axis (the Series index), bar height = wins. CSK and MI are the tallest. rot=0 keeps labels horizontal. Bar charts are the default choice for "compare categories".

Example — line chart over seasons

season_runs = pd.DataFrame(
    {"MI": [2380, 2410, 2650], "CSK": [2520, 2300, 2150]},
    index=[2018, 2019, 2020],
)
season_runs.plot(kind="line", marker="o", title="Total runs per season")
plt.show()

What you see: two lines (one per column) across 2018–2020, with a legend for MI and CSK. Use a line chart only when the x-axis has a natural order — never for unordered categories like team names.

Example — histogram and box plot for distributions

import numpy as np

rng = np.random.default_rng(42)
scores = pd.DataFrame({
    "MI": rng.normal(170, 20, 50).round(),
    "CSK": rng.normal(165, 12, 50).round(),
})
scores.plot(kind="hist", bins=15, alpha=0.6, title="First-innings totals")
scores.plot(kind="box", title="First-innings totals by team")
plt.show()

What you see: the histogram shows two overlapping "hills" — where scores are most common and how wide they spread (alpha=0.6 makes them see-through). The box plot shows each team's median (middle line), the middle 50% (the box), the whiskers, and dots for outliers. MI's taller box means its scores vary more.

Example — scatter for relationships, pie for parts of a whole

cars = pd.DataFrame({"horsepower": [89, 103, 114, 148, 201, 182, 130],
                     "price": [6.5, 9.5, 13.5, 19.5, 33.0, 21.0, 15.5]})
cars.plot(kind="scatter", x="horsepower", y="price", title="Horsepower vs price")

toss = pd.Series({"field": 15, "bat": 3})
toss.plot(kind="pie", autopct="%.1f%%", title="Toss decisions", ylabel="")
plt.show()

What you see: the scatter shows dots rising from bottom-left to top-right — the visual form of the strong positive correlation (0.95) found in the car project. The pie shows 83.3% field vs 16.7% bat. Use pies only for a handful of slices; with many categories a bar chart is easier to compare.

Typical notebook patterns

df["city"].value_counts().head(10).plot(kind="barh")           # top 10 categories
df["year"].value_counts().sort_index().plot(kind="bar")        # counts per year, in year order
df.groupby("brand")["price"].mean().sort_values().plot(kind="bar")  # average per group
df[["runs", "balls"]].plot(subplots=True, figsize=(8, 6))      # one chart per column

Common mistakes

• ❌ Plotting value_counts() of years without .sort_index() — bars appear in count order, not year order. ✅ .value_counts().sort_index().plot(...).
• ❌ A line chart for unordered categories (teams). ✅ Use bar/barh.
• ❌ A pie chart with 15 slices. ✅ Pies for ≤ 5 parts; otherwise barh.
• ❌ kind="scatter" without x and y. ✅ Scatter and hexbin need both column names.
• ❌ Nothing appears in a .py script. ✅ Call plt.show() (Jupyter shows charts automatically).
• ❌ Plotting a text column — "no numeric data to plot". ✅ Plot counts (value_counts()) or numeric columns.

Interview tip

Q: Which chart would you use to check whether a numeric column has outliers? A: A box plot (df["col"].plot(kind="box")) — points beyond the whiskers (1.5 × IQR from the box) are drawn individually as outliers. A histogram is the companion chart for seeing the overall shape of the distribution.