Series & DataFrames

ECON 3209 · Week 4, Lecture 1 · Kerala Agricultural University

Department of Development Economics, KAU

Autumn 2026

Learning Outcomes

By the end of this lecture, you will be able to:

1. Create pandas Series and DataFrame objects
2. Inspect rows, columns, and data types
3. Use loc and iloc for labelled and positional indexing
4. Add, modify, and summarise columns in a DataFrame
5. Connect tabular data handling to agricultural and banking applications

Why pandas?

• Real datasets usually have variable names and row labels
• NumPy arrays are fast, but they do not carry rich labels by default
• pandas adds labels, missing-data handling, and table operations
• It is the standard Python library for applied data wrangling
• Most econometrics workflows begin with a pandas DataFrame

Think of pandas as a programmable spreadsheet that is easier to audit, repeat, and scale.

Creating a Series

• A Series is a one-dimensional labelled object
• It contains values and an index
• It is useful for a single variable such as yield, price, or loan volume
• Methods like .mean() work directly on the data

Creating a DataFrame

• A DataFrame is a two-dimensional table
• Columns can represent different variables
• Each row is one observation such as a district or branch
• DataFrames are ideal for applied econometrics data preparation

Inspecting Structure

• .head() shows the first rows
• .info() reports column names, non-null counts, and data types
• Always inspect a dataset before analysis
• Early inspection prevents many downstream mistakes

Selecting Columns and Rows

• A single column selection returns a Series
• Double brackets return a smaller DataFrame
• Clear column selection is the first step in focused analysis
• Use explicit names instead of guessing positions whenever possible

loc for Label-Based Indexing

• loc selects by row and column labels
• It is useful when labels carry meaning, such as district names
• loc supports lists of labels and slices
• In applied work, label-based indexing is often the safest choice

iloc for Position-Based Indexing

• iloc selects by integer position
• It behaves more like NumPy slicing
• This is useful when you want “first two columns” or “rows 2 to 4”
• Be careful: positions can change if the DataFrame is rearranged

Creating New Columns

• New variables are easy to create from existing columns
• The calculation happens element-wise by row
• This is how we build indicators and ratios for analysis
• Derived columns should be clearly named and documented

Filtering Rows

• The condition creates a boolean mask
• The DataFrame keeps only rows where the mask is true
• This is a standard way to subset data before charts or models
• Filtering is more reproducible than manual spreadsheet deletion

Summary Statistics in pandas

• .describe() quickly summarises numeric columns
• It reports count, mean, standard deviation, and key percentiles
• This is a useful first diagnostic for any new dataset
• pandas integrates descriptive statistics with tabular structure

Exercise

Create a DataFrame with three districts and columns for deposit_crore and loan_crore.

1. Add a new column called gap equal to deposits minus loans.
2. Use loc to print the row for one district.
3. Filter rows where gap is greater than 20.

Summary

• ✅ pandas adds labels and table structure to numerical data
• ✅ Series handle one variable; DataFrame handles many variables together
• ✅ loc uses labels and iloc uses positions
• ✅ New columns can be created with simple vectorised expressions
• ✅ Filtering and summary statistics are built into the workflow
• ✅ pandas is the main entry point for real econometrics datasets

Next Lecture

Merge, Reshape & Groupby

• We will learn how to combine datasets and summarise groups
• These skills are essential for turning raw data into analysis-ready tables
• They also mirror common tasks in official statistics and banking data systems