head/tail & Statistics

Once you have data, you'll want to peek at it and summarise it. Pandas gives you one-liners for both — from head() to a full describe().

1Peeking with head() & tail()

Big tables are unwieldy — head(n) shows the first n rows and tail(n) the last. Both default to 5.

df_head_tail.py

import pandas as pd

df = pd.DataFrame({
    'Name': ['Rinku', 'Ritu', 'Pankaj', 'Ajay', 'Aditya', 'Vinay'],
    'IP':   [89, 83, 95, 99, 87, 98],
})

# Default case: no arguments returns the first 5 rows
print("df.head() [Default 5 rows]:")
print(df.head())

# Custom count: passing a number returns exactly that many rows
print("\n--- df.head(2) ---")
print(df.head(2))

# Default case: no arguments returns the last 5 rows
print("\n--- df.tail() [Default 5 rows] ---")
print(df.tail())

# Custom count: passing a number returns exactly that many rows
print("\n--- df.tail(2) ---")
print(df.tail(2))

import pandas as pd

df = pd.DataFrame({
    'Name': ['Rinku', 'Ritu', 'Pankaj', 'Ajay', 'Aditya', 'Vinay'],
    'IP':   [89, 83, 95, 99, 87, 98],
})

# Default case: no arguments returns the first 5 rows
print("df.head() [Default 5 rows]:")
print(df.head())

# Custom count: passing a number returns exactly that many rows
print("\n--- df.head(2) ---")
print(df.head(2))

# Default case: no arguments returns the last 5 rows
print("\n--- df.tail() [Default 5 rows] ---")
print(df.tail())

# Custom count: passing a number returns exactly that many rows
print("\n--- df.tail(2) ---")
print(df.tail(2))

2Summaries run along an axisOptional Reading

Functions like sum(), mean(), max() and min() collapse the table. The axis argument decides the direction:

Which way does it summarise?

Call	Direction	Answers
df.sum()	axis=0 (default) — down each column	one number per column
df.sum(axis=1)	across each row	one number per row

df_stats_axis.py

import pandas as pd
import numpy as np

production = pd.DataFrame({
    'Rice':   [7452.4, 1930.0, 2604.8, 11586.2],
    'Wheat':  [np.nan, 2737.0, np.nan, 16440.5],
    'Pulses': [931.0, 818.0, 1.7, 33.0],
}, index=['AP', 'Gujarat', 'Kerala', 'Punjab'])

# Default axis=0: total for EACH crop (down the columns)
print("Total per crop:")
print(production.sum())

# axis=1: total for EACH state (across the row)
print("\nTotal per state:")
print(production.sum(axis=1))

# max / min / mean work the same way
print("\nHighest per crop:")
print(production.max())

import pandas as pd
import numpy as np

production = pd.DataFrame({
    'Rice':   [7452.4, 1930.0, 2604.8, 11586.2],
    'Wheat':  [np.nan, 2737.0, np.nan, 16440.5],
    'Pulses': [931.0, 818.0, 1.7, 33.0],
}, index=['AP', 'Gujarat', 'Kerala', 'Punjab'])

# Default axis=0: total for EACH crop (down the columns)
print("Total per crop:")
print(production.sum())

# axis=1: total for EACH state (across the row)
print("\nTotal per state:")
print(production.sum(axis=1))

# max / min / mean work the same way
print("\nHighest per crop:")
print(production.max())

Note

These functions skip NaN by default — a missing value won't break your total. count() even tells you how many real (non-NaN) values each column has.

3The all-in-one: describe() & info()Optional Reading

describe() reports count, mean, std, min, the quartiles and max for every numeric column in one shot. info() summarises the structure — columns, non-null counts and dtypes.

df_describe.py

import pandas as pd
import numpy as np

production = pd.DataFrame({
    'Rice':   [7452.4, 1930.0, 2604.8, 11586.2],
    'Wheat':  [np.nan, 2737.0, np.nan, 16440.5],
    'Pulses': [931.0, 818.0, 1.7, 33.0],
}, index=['AP', 'Gujarat', 'Kerala', 'Punjab'])

print(production.describe())

print("\n--- info() ---")
production.info()

import pandas as pd
import numpy as np

production = pd.DataFrame({
    'Rice':   [7452.4, 1930.0, 2604.8, 11586.2],
    'Wheat':  [np.nan, 2737.0, np.nan, 16440.5],
    'Pulses': [931.0, 818.0, 1.7, 33.0],
}, index=['AP', 'Gujarat', 'Kerala', 'Punjab'])

print(production.describe())

print("\n--- info() ---")
production.info()

4Statistics on a subsetOptional Reading

The real trick: first select a row, column or subset (with the tools from the earlier lessons), then apply the function to just that piece.

df_stats_subset.py

import pandas as pd
import numpy as np

production = pd.DataFrame({
    'Rice':   [7452.4, 1930.0, 2604.8, 11586.2],
    'Wheat':  [np.nan, 2737.0, np.nan, 16440.5],
    'Pulses': [931.0, 818.0, 1.7, 33.0],
}, index=['AP', 'Gujarat', 'Kerala', 'Punjab'])

# Highest Rice production of all
print("Max Rice:", production['Rice'].max())

# Average across just Punjab's crops
print("Punjab mean:", production.loc['Punjab'].mean())

# Total Pulses across the first two states
print("Pulses (AP+Gujarat):", production['Pulses'].iloc[0:2].sum())

import pandas as pd
import numpy as np

production = pd.DataFrame({
    'Rice':   [7452.4, 1930.0, 2604.8, 11586.2],
    'Wheat':  [np.nan, 2737.0, np.nan, 16440.5],
    'Pulses': [931.0, 818.0, 1.7, 33.0],
}, index=['AP', 'Gujarat', 'Kerala', 'Punjab'])

# Highest Rice production of all
print("Max Rice:", production['Rice'].max())

# Average across just Punjab's crops
print("Punjab mean:", production.loc['Punjab'].mean())

# Total Pulses across the first two states
print("Pulses (AP+Gujarat):", production['Pulses'].iloc[0:2].sum())

Key Takeaway

Peek with head()/tail(). Summarise with sum/mean/max/min/count — axis=0 works down columns, axis=1 across rows — or get everything with describe(). Select first to summarise a subset.

Quick Check

production.sum(axis=1) gives you…

Quick Check

A column has one NaN. What does df['col'].sum() do with it?

← Previous

🔁 Iterating over a DataFrame

CSV Files 📄