Day 1: DataFrames — Loading, Inspecting, Selecting

What You'll Build Today

A data exploration script that loads a sales CSV, prints a full diagnostic summary (shape, dtypes, null counts, top values), then filters and exports two subsets: high-value orders and orders from a specific region.

Setup

Installing pandas

If you haven't already, install pandas in your Python environment:

bash

pip install pandas openpyxl

The openpyxl package lets pandas read Excel files. You'll need it on Day 1.

Check your version: Run python -c "import pandas as pd; print(pd.__version__)" — you want 1.5 or higher. 2.x is ideal.

Loading Data

Reading files into DataFrames

A DataFrame is a table — rows and columns — held in memory. You create one by reading a file.

pythonload.py

import pandas as pd

# CSV — the most common format
df = pd.read_csv("sales.csv")

# CSV with custom separator (semicolons, tabs, etc.)
df = pd.read_csv("data.csv", sep=";")

# Excel — specify the sheet name
df = pd.read_excel("report.xlsx", sheet_name="Sheet1")

# JSON — works for both arrays and nested objects
df = pd.read_json("orders.json")

# From a URL (no download needed)
url = "https://raw.githubusercontent.com/.../data.csv"
df = pd.read_csv(url)

All of these return the same thing: a DataFrame. The API is consistent no matter where the data comes from.

Quick tip: read_csv accepts file paths, URLs, and even file-like objects. If your data is behind an API, you can fetch it with requests and pass the content directly to pandas.

Inspection

Understanding what you loaded

Before you do anything with data, you need to know what you have. These five commands tell you everything:

pythoninspect.py

# Shape: (rows, columns)
print(df.shape)         # (10420, 8)

# First 5 rows
print(df.head())

# Column names, data types, and non-null counts
df.info()

# Numeric summary: mean, std, min, max, quartiles
print(df.describe())

# Count nulls per column
print(df.isnull().sum())

Run these on any new dataset before touching anything else. They tell you: how big is this? what types are the columns? where are the holes?

pythoninspect2.py

# Value counts for a categorical column
print(df["region"].value_counts())

# Unique values in a column
print(df["status"].unique())

# Data types of every column
print(df.dtypes)

# Percentage of nulls per column
print(df.isnull().mean() * 100)

Selection

loc, iloc, and boolean filtering

Three ways to pull data from a DataFrame. Each has a specific use case.

Column selection

python

# Single column → returns a Series
revenue = df["revenue"]

# Multiple columns → returns a DataFrame
subset = df[["order_id", "revenue", "region"]]

loc — label-based selection

python

# loc[row_label, column_label]
# Select rows 0–4, all columns
df.loc[0:4]

# Select specific rows and columns by name
df.loc[0:4, ["order_id", "revenue"]]

# Select a single cell
df.loc[5, "revenue"]

iloc — position-based selection

python

# iloc[row_position, column_position] — uses integers
df.iloc[0:5]           # first 5 rows
df.iloc[:, 0:3]         # all rows, first 3 columns
df.iloc[-5:]            # last 5 rows

Boolean filtering — the one you'll use most

python

# Rows where revenue > 1000
high_value = df[df["revenue"] > 1000]

# Multiple conditions — use & (and) and | (or)
west_high = df[(df["region"] == "West") & (df["revenue"] > 500)]

# Filter by list of values
key_regions = df[df["region"].isin(["West", "East"])]

# Filter rows where a column contains a string
returns = df[df["status"].str.contains("returned", case=False)]

Common mistake: When combining conditions, always wrap each condition in parentheses. df[a == 1 & b == 2] will give a cryptic error. df[(a == 1) & (b == 2)] works.

Practice Challenge

Build a data diagnostic script

Download any CSV from Kaggle or use your own data. Write a script that:

Loads the file and prints shape, column names, and dtypes
Prints the count and percentage of nulls for every column
Filters to rows where a numeric column is above its mean
Saves those rows to a new CSV with df.to_csv("output.csv", index=False)

What You Learned Today

Load CSV, Excel, and JSON files into DataFrames
Inspect shape, dtypes, nulls, and value distributions
Select columns by name with bracket notation
Use loc (label-based) and iloc (position-based) indexing
Filter rows with boolean masks and isin()

Course Progress

Day 1 of 5 — 20%

Day 1 Complete

Day 2: Cleaning Data

Tomorrow you'll handle real-world messy data: missing values, duplicates, and wrong data types.

Start Day 2

← Course Overview Join the Bootcamp