Introduction to pandas: Series and DataFrames

pandas is the most important library for working with tabular data in Python — think of it as a programmable spreadsheet. It reads data from CSV, Excel, JSON, SQL databases and more into a DataFrame: a table with labelled columns (each with its own type) and a labelled row index. It then lets you inspect, clean, filter, transform, group, join and export that data in a few lines.

This lesson introduces the two core objects, Series and DataFrame, shows how to create them and load files, and how to take a first look at a new dataset. The examples use pandas 3, which enables Copy-on-Write and a dedicated string type by default.

Series and DataFrame

A Series is a one-dimensional labelled array — one column — with an index (labels), values and a dtype. A DataFrame is a collection of Series sharing the same index. Create one from a dict of lists (column by column), a list of dicts (row by row, like JSON records), or a NumPy array with column names. Selecting df["price"] returns a Series; df[["name", "price"]] returns a smaller DataFrame. Install with pip install pandas and import as import pandas as pd.

Reading and Writing Data

pd.read_csv("sales.csv") — options include sep, usecols, dtype, parse_dates, index_col, na_values and nrows.
pd.read_excel("report.xlsx", sheet_name="2026") (needs openpyxl), pd.read_json(), pd.read_sql(query, connection), pd.read_parquet().
df.to_csv("out.csv", index=False), to_excel, to_json(orient="records"), to_sql, to_parquet.

First Look at a Dataset

head(n)/tail(n) show the first/last rows, shape gives (rows, columns), columns and dtypes list columns and types, info() summarises types, non-null counts and memory, describe() gives count, mean, std, min, quartiles and max for numeric columns, and value_counts() counts categories. Doing this first on every new dataset reveals wrong types, missing values and surprises before you analyse anything.

Examples

Creating Series and DataFrames

Python

import pandas as pd

prices = pd.Series([450, 899, 350], index=["Python Basics", "Deep Python", "SQL Basics"], name="price")
print(prices)
print(prices["Deep Python"], prices.mean(), prices.index.tolist())

# From a dict of columns
df = pd.DataFrame({
    "product": ["Pen", "Notebook", "Backpack", "Mouse"],
    "category": ["stationery", "stationery", "bags", "electronics"],
    "price": [20, 60, 1250, 699],
    "in_stock": [True, True, False, True],
})
print(df)
print(df.shape, list(df.columns))
print(df.dtypes)

# From a list of dicts (like JSON records)
orders = pd.DataFrame([{"id": 1, "total": 540.5}, {"id": 2, "total": 120.0}])
print(orders)

Output

Python Basics    450
Deep Python      899
SQL Basics       350
Name: price, dtype: int64
899 566.3333333333334 ['Python Basics', 'Deep Python', 'SQL Basics']
    product     category  price  in_stock
0       Pen   stationery     20      True
1  Notebook   stationery     60      True
2  Backpack         bags   1250     False
3     Mouse  electronics    699      True
(4, 4) ['product', 'category', 'price', 'in_stock']
product       str
category      str
price       int64
in_stock     bool
dtype: object
   id  total
0   1  540.5
1   2  120.0

Reading a CSV and taking a first look

Python

import io
import pandas as pd

csv_text = """order_id,date,city,product,quantity,unit_price
1001,2026-01-05,Pune,Pen,10,20
1002,2026-01-05,Mumbai,Notebook,4,60
1003,2026-01-06,Pune,Backpack,1,1250
1004,2026-01-07,Delhi,Mouse,2,699
1005,2026-01-07,Mumbai,Pen,25,20
1006,2026-01-08,Pune,Notebook,,60
"""
# With a real file: pd.read_csv("sales.csv", parse_dates=["date"])
sales = pd.read_csv(io.StringIO(csv_text), parse_dates=["date"])

print(sales.head(3))
print(sales.shape)
sales.info()
print(sales.describe().round(1))
print(sales["city"].value_counts())

Output

   order_id       date    city   product  quantity  unit_price
0      1001 2026-01-05    Pune       Pen      10.0          20
1      1002 2026-01-05  Mumbai  Notebook       4.0          60
2      1003 2026-01-06    Pune  Backpack       1.0        1250
(6, 6)
<class 'pandas.DataFrame'>
RangeIndex: 6 entries, 0 to 5
Data columns (total 6 columns):
 #   Column      Non-Null Count  Dtype
---  ------      --------------  -----
 0   order_id    6 non-null      int64
 1   date        6 non-null      datetime64[us]
 2   city        6 non-null      str
 3   product     6 non-null      str
 4   quantity    5 non-null      float64
 5   unit_price  6 non-null      int64
dtypes: datetime64[us](1), float64(1), int64(2), str(2)
memory usage: 420.0 bytes
       order_id                 date  quantity  unit_price
count       6.0                    6       5.0         6.0
mean     1003.5  2026-01-06 08:00:00       8.4       351.5
min      1001.0  2026-01-05 00:00:00       1.0        20.0
25%      1002.2  2026-01-05 06:00:00       2.0        30.0
50%      1003.5  2026-01-06 12:00:00       4.0        60.0
75%      1004.8  2026-01-07 00:00:00      10.0       539.2
max      1006.0  2026-01-08 00:00:00      25.0      1250.0
std         1.9                  NaN       9.9       513.4
city
Pune      3
Mumbai    2
Delhi     1
Name: count, dtype: int64

Writing data back out

Python

import pandas as pd

report = pd.DataFrame({"city": ["Pune", "Mumbai"], "revenue": [1890, 740]})
report.to_csv("report.csv", index=False)           # index=False: no extra index column
print(open("report.csv").read())
print(report.to_json(orient="records"))
print(pd.read_csv("report.csv").equals(report))

Output

city,revenue
Pune,1890
Mumbai,740

[{"city":"Pune","revenue":1890},{"city":"Mumbai","revenue":740}]
True

Common Mistakes

Forgetting index=False in to_csv and getting an unwanted "Unnamed: 0" column when reading it back.
Not using parse_dates, so dates stay as text and date operations fail.
Skipping info()/describe() and missing wrong types or missing values.
Confusing df["col"] (a Series) with df[["col"]] (a DataFrame).
Loading an entire huge file when usecols or nrows would do.

Key Points to Remember

A Series is one labelled column; a DataFrame is a table of Series sharing an index.
Create DataFrames from dicts of lists, lists of dicts or NumPy arrays.
read_csv/read_excel/read_json/read_sql load data; to_csv and friends save it.
head, shape, dtypes, info, describe and value_counts give a first look.
pandas 3 uses Copy-on-Write and a dedicated string dtype by default.

Practice the examples

Change an input, predict the result, then compare it with the output. Explain why the result changes.

Use your local project environment for these examples. Codelab currently runs Python and HTML/CSS/JavaScript; framework examples may need project dependencies.