Reading CSV, Excel & JSON Files

import pandas as pd

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

# Common parameters:
df = pd.read_csv("sales.csv",
    sep=",",           # delimiter (use "\t" for TSV)
    header=0,          # row to use as column names
    skiprows=2,        # skip first 2 rows
    nrows=1000,        # read only 1000 rows
    usecols=["name","sales"],  # only these columns
    encoding="utf-8",  # or "latin-1" for Indian data
    na_values=["N/A", "-", "NULL"]  # treat as NaN
)

encoding="latin-1" (or "cp1252") is the most common fix for Indian data files exported from Excel. na_values lets you define what counts as missing beyond the default NaN.

df = pd.read_excel("report.xlsx")

# Specific sheet:
df = pd.read_excel("report.xlsx", sheet_name="Sales")

# Multiple sheets → dict of DataFrames:
sheets = pd.read_excel("report.xlsx", sheet_name=None)
df_sales = sheets["Sales"]
df_hr = sheets["HR"]

# Skip rows and specific columns:
df = pd.read_excel("report.xlsx", skiprows=3, usecols="A:E")

sheet_name=None reads all sheets at once. This is useful when you receive monthly reports where each month is a separate sheet and you need to combine them.

# From file:
df = pd.read_json("data.json")

# From API response (list of dicts):
import json, requests
response = requests.get("https://api.example.com/data")
data = response.json()    # list of dicts
df = pd.DataFrame(data)

# Nested JSON — normalize:
from pandas import json_normalize
df = json_normalize(data, record_path="items")

json_normalize() is essential for nested JSON — the kind you get from REST APIs where each record has sub-objects. Without it, nested fields come in as dictionaries inside cells.

# CSV:
df.to_csv("output.csv", index=False)  # index=False prevents saving row numbers

# Excel:
df.to_excel("output.xlsx", index=False, sheet_name="Results")

# Multiple sheets:
with pd.ExcelWriter("report.xlsx") as writer:
    df_sales.to_excel(writer, sheet_name="Sales", index=False)
    df_hr.to_excel(writer, sheet_name="HR", index=False)

Always use index=False unless the index contains meaningful data. ExcelWriter is the correct way to write multiple sheets — do not call to_excel() on the same file twice.

# Read in chunks:
for chunk in pd.read_csv("large.csv", chunksize=10000):
    process(chunk)  # process each 10k row chunk

# Only load needed columns:
df = pd.read_csv("large.csv", usecols=["id","sales","date"])

# Reduce memory with dtype specification:
df = pd.read_csv("large.csv", dtype={"id": "int32", "sales": "float32"})

For files larger than 1GB, chunking is necessary. Specifying dtypes upfront can reduce memory usage by 50-75% — int32 vs int64, float32 vs float64, and category for repeated strings.