Pandas Tutorial: 30 Must-Know Functions With

Introduction

Pandas is the workhorse of data analysis in Python. For Cinute Digital Pvt Ltd (CDPL) learners and partner teams, this tutorial covers the 30 functions that deliver most day to day value. Each snippet uses realistic mini datasets so you can copy, run, and adapt to your projects.

Tip: Keep a notebook open and paste each block to build your own cheat sheet.

Setup and Sample Data

import pandas as pd

# Orders dataset
orders = pd.DataFrame({
    "order_id":[101,102,103,104,105],
    "user_id":[1,1,2,3,2],
    "country":["IN","IN","AE","IN","AE"],
    "amount":[1200,800,560,1500,300],
    "order_date": pd.to_datetime(["2025-10-01","2025-10-02","2025-10-02","2025-10-03","2025-10-03"])
})

# Users dataset
users = pd.DataFrame({
    "user_id":[1,2,3],
    "name":["Asha","Bilal","Chirag"],
    "segment":["pro","basic","pro"]
})

Two small but realistic tables we will reuse through the tutorial

Import, Inspect, and Save

read_csv load a CSV

df = pd.read_csv("sales.csv")  # delimiter, dtype, parse_dates are useful args

to_csv save a CSV

orders.to_csv("orders_out.csv", index=False)

head and tail quick peek

orders.head(3); orders.tail(2)

info schema and nulls

orders.info()

describe stats summary

orders.describe(numeric_only=True)

Selecting and Filtering

loc label based select

orders.loc[orders["country"]=="IN", ["order_id","amount"]]

iloc position based select

orders.iloc[:3, :2]

query readable filters

orders.query("amount >= 800 and country == 'IN'")

isin membership

orders[orders["country"].isin(["IN","AE"])]

between numeric ranges

orders[orders["amount"].between(500,1200)]

Transformation and Features

assign create columns

aug = orders.assign(gst=lambda d: d.amount * 0.18, total=lambda d: d.amount * 1.18)

astype convert types

orders["user_id"] = orders["user_id"].astype("int64")

replace value mapping

orders["country"] = orders["country"].replace({"IN":"India","AE":"UAE"})

cut and qcut binning

orders["amount_bucket"] = pd.qcut(orders["amount"], q=3, labels=["low","med","high"])

apply row or column wise logic

orders["flag_big"] = orders["amount"].apply(lambda x: x >= 1000)

Missing Data and Cleaning

isna and notna

orders.isna().sum()

fillna impute values

orders["amount"] = orders["amount"].fillna(orders["amount"].median())

dropna remove null rows

orders = orders.dropna(subset=["amount"])

drop_duplicates deduplicate

orders = orders.drop_duplicates(["order_id"])

rename readable columns

orders = orders.rename(columns={"order_date":"date"})

Join, Group, and Pivot

merge relational join

joined = orders.merge(users, on="user_id", how="left")

groupby aggregate by keys

joined.groupby("country")["amount"].agg(["count","mean","sum"]).reset_index()

pivot_table quick reporting

pd.pivot_table(joined, values="amount", index="country", columns="segment", aggfunc="sum", fill_value=0)

melt unpivot for tidy data

wide = pd.DataFrame({"city":["Mumbai","Dubai"],"Q1":[10,8],"Q2":[12,11]})
tidy = wide.melt(id_vars="city", var_name="quarter", value_name="sales")

crosstab frequency table

pd.crosstab(joined["country"], joined["segment"])

Dates, Windows, and Sorting

to_datetime parse dates

orders["date"] = pd.to_datetime(orders["order_date"])

dt accessors for features

orders["dow"] = orders["date"].dt.day_name()

sort_values ordering

orders.sort_values(["amount","date"], ascending=[False, True])

rolling window metrics

orders.sort_values("date", inplace=True)
orders["rolling_total"] = orders["amount"].rolling(window=2).sum()

nlargest and nsmallest top k

orders.nlargest(3, "amount")[["order_id","amount"]]

Mini Project: Daily Revenue by Country and Segment

Combine several functions to answer a realistic question.

report = (
    orders.merge(users, on="user_id", how="left")
          .assign(date=lambda d: pd.to_datetime(d["order_date"]))
          .groupby(["date","country","segment"], as_index=False)["amount"]
          .sum()
)

# Pivot to a dashboard friendly view
dashboard = pd.pivot_table(
    report, values="amount",
    index=["date","country"], columns="segment",
    aggfunc="sum", fill_value=0
).reset_index()

print(dashboard.head())

From raw orders to a tidy report you can chart in seconds

Common Pitfalls and Quick Fixes

Wrong dtype for ids or codes use string types to keep leading zeros.
Silent date parsing issues always run pd.to_datetime with format if known.
Unexpected duplicates on merge check key uniqueness with value_counts before joining.
Groupby returning index use as_index=False or reset_index for flat tables.
Chained assignment warnings prefer .loc for explicit assignment.

Practice Datasets You Can Try

Orders and users from your project database export.
Retail sales by month from a public government portal.
Website events CSV from analytics tools.
Helpdesk tickets with status and resolution time.

Create a folder with raw, interim, and processed CSVs to mirror a simple data workflow.

Conclusion

You now have a compact toolkit of 30 Pandas functions that handle most everyday analysis. Keep this page as a reference, and turn the mini project into a dashboard or a report. Consistent practice with real data is the fastest way to mastery.

Introduction

Tip: Keep a notebook open and paste each block to build your own cheat sheet.

Setup and Sample Data

import pandas as pd

# Orders dataset
orders = pd.DataFrame({
    "order_id":[101,102,103,104,105],
    "user_id":[1,1,2,3,2],
    "country":["IN","IN","AE","IN","AE"],
    "amount":[1200,800,560,1500,300],
    "order_date": pd.to_datetime(["2025-10-01","2025-10-02","2025-10-02","2025-10-03","2025-10-03"])
})

# Users dataset
users = pd.DataFrame({
    "user_id":[1,2,3],
    "name":["Asha","Bilal","Chirag"],
    "segment":["pro","basic","pro"]
})

Two small but realistic tables we will reuse through the tutorial

Import, Inspect, and Save

read_csv load a CSV

df = pd.read_csv("sales.csv")  # delimiter, dtype, parse_dates are useful args

to_csv save a CSV

orders.to_csv("orders_out.csv", index=False)

head and tail quick peek

orders.head(3); orders.tail(2)

info schema and nulls

orders.info()

describe stats summary

orders.describe(numeric_only=True)

Selecting and Filtering

loc label based select

orders.loc[orders["country"]=="IN", ["order_id","amount"]]

iloc position based select

orders.iloc[:3, :2]

query readable filters

orders.query("amount >= 800 and country == 'IN'")

isin membership

orders[orders["country"].isin(["IN","AE"])]

between numeric ranges

orders[orders["amount"].between(500,1200)]

Transformation and Features

assign create columns

aug = orders.assign(gst=lambda d: d.amount * 0.18, total=lambda d: d.amount * 1.18)

astype convert types

orders["user_id"] = orders["user_id"].astype("int64")

replace value mapping

orders["country"] = orders["country"].replace({"IN":"India","AE":"UAE"})

cut and qcut binning

orders["amount_bucket"] = pd.qcut(orders["amount"], q=3, labels=["low","med","high"])

apply row or column wise logic

orders["flag_big"] = orders["amount"].apply(lambda x: x >= 1000)

Missing Data and Cleaning

isna and notna

orders.isna().sum()

fillna impute values

orders["amount"] = orders["amount"].fillna(orders["amount"].median())

dropna remove null rows

orders = orders.dropna(subset=["amount"])

drop_duplicates deduplicate

orders = orders.drop_duplicates(["order_id"])

rename readable columns

orders = orders.rename(columns={"order_date":"date"})

Join, Group, and Pivot

merge relational join

joined = orders.merge(users, on="user_id", how="left")

groupby aggregate by keys

joined.groupby("country")["amount"].agg(["count","mean","sum"]).reset_index()

pivot_table quick reporting

pd.pivot_table(joined, values="amount", index="country", columns="segment", aggfunc="sum", fill_value=0)

melt unpivot for tidy data

wide = pd.DataFrame({"city":["Mumbai","Dubai"],"Q1":[10,8],"Q2":[12,11]})
tidy = wide.melt(id_vars="city", var_name="quarter", value_name="sales")

crosstab frequency table

pd.crosstab(joined["country"], joined["segment"])

Dates, Windows, and Sorting

to_datetime parse dates

orders["date"] = pd.to_datetime(orders["order_date"])

dt accessors for features

orders["dow"] = orders["date"].dt.day_name()

sort_values ordering

orders.sort_values(["amount","date"], ascending=[False, True])

rolling window metrics

orders.sort_values("date", inplace=True)
orders["rolling_total"] = orders["amount"].rolling(window=2).sum()

nlargest and nsmallest top k

orders.nlargest(3, "amount")[["order_id","amount"]]

Mini Project: Daily Revenue by Country and Segment

Combine several functions to answer a realistic question.

report = (
    orders.merge(users, on="user_id", how="left")
          .assign(date=lambda d: pd.to_datetime(d["order_date"]))
          .groupby(["date","country","segment"], as_index=False)["amount"]
          .sum()
)

# Pivot to a dashboard friendly view
dashboard = pd.pivot_table(
    report, values="amount",
    index=["date","country"], columns="segment",
    aggfunc="sum", fill_value=0
).reset_index()

print(dashboard.head())

From raw orders to a tidy report you can chart in seconds

Common Pitfalls and Quick Fixes

Wrong dtype for ids or codes use string types to keep leading zeros.
Silent date parsing issues always run pd.to_datetime with format if known.
Unexpected duplicates on merge check key uniqueness with value_counts before joining.
Groupby returning index use as_index=False or reset_index for flat tables.
Chained assignment warnings prefer .loc for explicit assignment.

Practice Datasets You Can Try

Orders and users from your project database export.
Retail sales by month from a public government portal.
Website events CSV from analytics tools.
Helpdesk tickets with status and resolution time.

Create a folder with raw, interim, and processed CSVs to mirror a simple data workflow.

Pandas Tutorial: 30 Must-Know Functions With Real Datasets

Introduction

Setup and Sample Data

Import, Inspect, and Save

Selecting and Filtering

Transformation and Features

Missing Data and Cleaning

Join, Group, and Pivot

Dates, Windows, and Sorting

Mini Project: Daily Revenue by Country and Segment

Common Pitfalls and Quick Fixes

Practice Datasets You Can Try

Conclusion

Tags

Share this article

Pandas Tutorial: 30 Must-Know Functions With Real Datasets

Introduction

Setup and Sample Data

Import, Inspect, and Save

Selecting and Filtering

Transformation and Features

Missing Data and Cleaning

Join, Group, and Pivot

Dates, Windows, and Sorting

Mini Project: Daily Revenue by Country and Segment

Common Pitfalls and Quick Fixes

Practice Datasets You Can Try

Conclusion

Tags

Share this article

Ready for Career Guidance?