Load, inspect, and save data
Read tabular data and perform initial checks before analyzing it.
Read a CSV file
read_csv loads comma-separated tabular data into a DataFrame. Use a small in-memory CSV here so the example can run without a separate data file:
from io import StringIO
import pandas as pd
csv_text = """product,region,units
tea,west,12
coffee,east,8
juice,west,5
"""
sales = pd.read_csv(StringIO(csv_text))
print(sales.head(2))
Pandas has readers and writers for a variety of formats. Additional engines or packages may be required for some formats, such as Excel workbooks. Use options such as parse_dates or dtype when you know how a column should be interpreted, and verify the result after loading.
Inspect before changing the data
Check the size, column names, data types, and a preview to catch parsing surprises early:
print(sales.shape) # (3, 3)
print(sales.columns) # column labels
print(sales.dtypes) # one dtype per column
print(sales.info()) # non-null counts and memory summary
print(sales.describe()) # summary for numeric columns
shape and dtypes are attributes, so they do not have parentheses. head, info, and describe are methods, so they do. info() prints its report and returns None.
Save a result
Write a DataFrame to CSV with to_csv. Set index=False if the row index is not part of the data you want in the file:
output = StringIO()
sales.to_csv(output, index=False)
print(output.getvalue().splitlines()[0]) # product,region,units
Exercise: Change the input CSV to include one more row. Load it, check the shape and dtypes, and write it back without the generated row index.
Sources: pandas I/O tools, pandas introductory tutorial: read and write tabular data, GeeksforGeeks: Pandas Introduction