Kaggle
CC BY 4.0
Crime Headline Classification
A balanced binary-classification set of 7,117 news headlines labelled crime or non-crime.
What it contains
What this is
A balanced binary-classification dataset of 7,117 news headlines, each labelled as crime-related or not. It is a clean starting point for news classification, adapted from the News Category Dataset by Rishabh Misra.
| Headlines | 7,117 |
| Classes | crime (1) and non-crime (0), balanced |
| Format | One CSV, 205 kB |
| Views | 3,600+ |
| Downloads | 540 |
| Licence | CC BY 4.0 |
Columns
| Column | Type | Meaning |
|---|---|---|
| title | text | The news headline |
| is_crime_report | 0 or 1 | Whether the headline reports a crime |
Use it for
- Binary text classification, as a teaching or baseline task
- Crime-news detection and filtering
- News categorisation pipelines
It came out of my work on news scraping and crime-data pipelines, including the Crime Scraper project.
Headlines
7,117
Balance
50 / 50
Downloads
540
Upvotes
15