Reading and Writing Data

Real data lives in files: CSV exports, Excel sheets, database dumps. Pandas reads all of them into a DataFrame with one function call.

Reading a CSV file#

A CSV (comma-separated values) file is plain text, one row per line:

date,product,units,price
2026-01-03,Pen,10,15
2026-01-03,Notebook,4,60
2026-01-04,Pen,,15
2026-01-05,Bag,1,850

If it’s saved as sales.csv next to your program, reading it is one line:

df = pd.read_csv("sales.csv")

To keep this lesson runnable anywhere, we’ll read the same text from memory with io.StringIO. It behaves exactly like a file:

import io
import pandas as pd

csv_text = """date,product,units,price
2026-01-03,Pen,10,15
2026-01-03,Notebook,4,60
2026-01-04,Pen,,15
2026-01-05,Bag,1,850"""

df = pd.read_csv(io.StringIO(csv_text))
print(df)

Output:

         date   product  units  price
0  2026-01-03       Pen   10.0     15
1  2026-01-03  Notebook    4.0     60
2  2026-01-04       Pen    NaN     15
3  2026-01-05       Bag    1.0    850

The empty units cell became NaN (not a number): Pandas’ marker for a missing value. Because NaN is a float, the whole units column turned into floats (10.0).

info(): the health check#

df.info()

Output:

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 4 entries, 0 to 3
Data columns (total 4 columns):
 #   Column   Non-Null Count  Dtype  
---  ------   --------------  -----  
 0   date     4 non-null      object 
 1   product  4 non-null      object 
 2   units    3 non-null      float64
 3   price    4 non-null      int64  
dtypes: float64(1), int64(1), object(2)
memory usage: 260.0+ bytes

info() tells you the number of rows, every column’s type, and how many values are non-null (not missing). units has 3 of 4, so one value is missing. Notice date is object (text), not a real date yet.

Useful read_csv options#

df = pd.read_csv(
    io.StringIO(csv_text),
    parse_dates=["date"],      # turn text into real dates
    usecols=["date", "product", "units"],  # load only some columns
)
print(df.dtypes)

Output:

date       datetime64[ns]
product            object
units             float64
dtype: object

Other options you’ll meet: sep=";" for semicolon files, nrows=1000 to load only the first rows of a huge file, encoding="latin-1" when you see strange characters.

Excel and other formats#

pd.read_excel("report.xlsx", sheet_name="Sales")   # needs: pip install openpyxl
pd.read_json("data.json")
pd.read_sql("SELECT * FROM orders", connection)

Saving your work#

df.to_csv("clean_sales.csv", index=False)
df.to_excel("clean_sales.xlsx", index=False)

index=False stops Pandas from writing the 0, 1, 2… row numbers as an extra column.

Quick check

A column that should be numbers shows object in info(). Why? Usually one cell contains text like "N/A" or "1,200". You’ll fix that in the cleaning lesson.

Summary#

• pd.read_csv("file.csv") loads a CSV; read_excel, read_json, read_sql load other formats.
• Empty cells become NaN.
• df.info() shows types and missing counts; run it on every new file.
• df.to_csv("out.csv", index=False) saves your result.