Find data. Ask better questions. Build better projects.

The largest community library of datasets, each with public notebooks so you can see how others analysed the same data.

Best forBeginners who want clean data plus example notebooks to learn from.

• Huge variety on every topic
• Community notebooks show real analysis
• Built-in free coding environment

Free account required

Each dataset has its own licence — check before commercial use.

IdeasRetail sales dashboard · Customer churn model

Classic, well-documented datasets used in ML research and university teaching for decades.

Best forMachine learning practice with small, clean, well-documented data.

• No sign-up — download instantly
• Every dataset is documented
• Standard benchmarks for comparing models

No account needed

Mostly CC BY 4.0 — confirm on each dataset page.

IdeasCredit risk classifier · Wine quality regression

A search engine that finds datasets published anywhere on the web — ideal when you need something niche.

Best forHunting down a specific topic nobody else in your class will have.

• Searches millions of datasets across the web
• Filters by format and usage rights
• Great for unique portfolio topics

No account needed

Depends on the original publisher.

IdeasFind niche data for a unique portfolio project

Official census, labour force, price and trade statistics for Pakistan — the source behind national headlines.

Best forPakistan-focused projects that stand out to local employers.

• Official national statistics
• Highly relevant for Pakistani employers
• Census, prices and labour data in one place

No account needed

Government publication — cite the source.

IdeasInflation trend dashboard · Provincial labour force analysis

Community open-data portal with ready-to-use datasets about Pakistan's districts, schools and infrastructure.

Best forClean Pakistan data in CSV without digging through PDF reports.

• Ready-to-use CSV files
• District-level detail
• Easy starting point for local projects

No account needed

Varies — check each dataset.

IdeasSchool enrolment analysis by district

Development indicators for every country — GDP, population, education, health — with a free API for engineers.

Best forCountry comparisons and learning to pull data from a real API.

• Consistent data for 200+ countries
• Free, documented API
• Decades of historical time series

No account needed

CC BY 4.0 for most indicators.

IdeasCompare South Asian economies · API-to-database pipeline

Research-quality data on health, energy, climate and living standards, maintained by Oxford researchers.

Best forStorytelling projects with trustworthy, citable global data.

• Extremely well documented
• Charts you can study for design ideas
• Perfect for long-term trend stories

No account needed

CC BY — credit the source.

IdeasEnergy transition storytelling report

The US government's open-data portal — hundreds of thousands of datasets on transport, health, crime and more.

Best forLarge, real-world messy datasets that test your cleaning skills.

• Enormous catalogue
• Many datasets have APIs
• Real government data — messy like real jobs

No account needed

Mostly public domain — confirm per dataset.

IdeasTraffic accident analysis

Earth science, climate and space mission data straight from NASA's programmes.

Best forAdvanced climate and time-series projects with serious wow factor.

• Unique space and climate data
• Well-maintained APIs
• Impressive portfolio material

No account needed

Generally public domain — confirm per dataset.

IdeasTemperature anomaly forecasting

Thousands of text, image and audio datasets for machine learning, loadable in two lines of Python.

Best forNLP and modern ML projects beyond plain tables.

• One-line loading in Python
• Text, image and audio — not just CSV
• The standard library for AI projects

Varies by dataset

Licence listed on each dataset card.

IdeasUrdu/English sentiment classifier

Large public datasets hosted on cloud storage — satellite imagery, genomics, climate — ideal for big-data practice.

Best forPractising Spark and cloud tools on genuinely large data.

• Multi-GB datasets for big-data tools
• Direct cloud access
• Real production-scale practice

No account needed

Varies by dataset.

IdeasSpark processing of multi-GB data

Clean, business-style datasets made specifically for portfolio dashboards, with guided project briefs.

Best forYour first Power BI or Excel portfolio dashboard.

• Designed for portfolio projects
• Clean data — focus on analysis, not cleaning
• Comes with project briefs to follow

No account needed

Free for learning and portfolio use — see site terms.

IdeasHR attrition dashboard · Restaurant orders analysis