Week 3 Slide Deck

Wrangling, Filtering; Formats

Jack Bandy 2026

Data Formats

How Do We Store This Data?

Spreadsheet??

You’ve probably seen this
Grid with cells, rows, columns (and sheets/tabs)
Excel, Google Sheets, Numbers, and LibreOffice Calc
Convenient for manual entry, review, sharing
But don’t keep your data in a spreadsheet…

	A	B	C	D	E
1	name	company	street	city	phone
2	Tyler Durden	Paper Street Soap Co.	537 Paper Street	Bradford	(288) 555-0153

CSV

Comma-separated values
Plain-text table: one row per record, one delimiter between fields
Often used for spreadsheets, exports, and simple datasets
Easy to inspect
types and hierarchy are usually implicit

name,company,product,street,city,postalCode,phone
Tyler Durden,Paper Street Soap Co.,All Natural Handmade,537 Paper Street,Bradford,19808,(288) 555-0153

TSV

Tab-separated values
Plain-text table like CSV, but fields are separated by tabs
Still flat: hierarchy and data types need outside context

name    company product street  city    postalCode  phone
Tyler Durden    Paper Street Soap Co.   All Natural Handmade    537 Paper Street    Bradford    19808   (288) 555-0153

JSON

JavaScript Object Notation
Text format for data interchange
Built from objects, arrays, strings, numbers, booleans, and null
Common for web APIs and configuration files

{
  "name": "Tyler Durden",
  "company": "Paper Street Soap Co.",
  "product": "All Natural Handmade",
  "address": {
    "street": "537 Paper Street",
    "city": "Bradford",
    "postalCode": "19808"
  },
  "phone": "(288) 555-0153"
}

XML

Extensible Markup Language
Nested tags represent elements and attributes
Verbose, but widely used by older document systems

<?xml version="1.0" encoding="UTF-8"?>
<person>
  <name>Tyler Durden</name>
  <company>Paper Street Soap Co.</company>
  <product>All Natural Handmade</product>
  <address>
    <street>537 Paper Street</street>
    <city>Bradford</city>
    <postalCode>19808</postalCode>
  </address>
  <phone>(288) 555-0153</phone>
</person>

YAML

YAML Ain’t Markup Language
Human-readable format based on indentation
Supports mappings, lists, scalars, and comments
Common for configuration files and data pipelines

name: Tyler Durden
company: Paper Street Soap Co.
product: All Natural Handmade
address:
  street: 537 Paper Street
  city: Bradford
  postalCode: 19808
phone: (288) 555-0153

Parquet

Binary columnar storage format
Efficient for large datasets
Stores schema and data types with the data
Common in Spark, DuckDB, Polars, and “data lakes”

message business_card {
  required binary name (STRING);
  required binary company (STRING);
  required binary product (STRING);
  required binary street (STRING);
  required binary city (STRING);
  required binary postalCode (STRING);
  required binary phone (STRING);
}

row 1:
Tyler Durden | Paper Street Soap Co. | All Natural Handmade |
537 Paper Street | Bradford | 19808 | (288) 555-0153

Format Summary

Format	Best fit	Short example
CSV	Flat tables, spreadsheet exports, simple datasets	`name,company,phone` `Tyler Durden,Paper Street Soap Co.,(288) 555-0153`
TSV	Flat text tables where commas may appear in fields	`name company phone`
JSON	APIs, nested records, web data	`{"name":"Tyler Durden","city":"Bradford"}`
XML	Document-like data with tags and attributes	`<name>Tyler Durden</name>`
YAML	Human-edited configuration and pipeline settings	`name: Tyler Durden` `city: Bradford`
Parquet	Typed, compressed analytics data	`name: STRING` `city: STRING`

Sources

GitHub source: https://github.com/jackbandy/data-science-fun/blob/main/docs/slides/week3.md.
Slide materials adapted from Elena Zheleva and Gonzalo Bello Lander.
Slide deck built with Quarto revealjs.
Format examples adapted from Wikipedia and project documentation: JSON, XML, Comma-separated values, YAML, Tab-separated values, Spreadsheet, and Apache Parquet.
Tyler Durden business card image: Wikimedia Commons remake by Michaelpreid, modified, CC BY-SA 4.0, https://commons.wikimedia.org/wiki/File:Tyler_Durden_Business_Card.png.
Title font is Big Shoulders; Body font is Libre Franklin.