Where can a Data Analyst find good, free public datasets for a portfolio project?

Absolutely. For a Data Analyst portfolio, I’d prioritize datasets that let you demonstrate more than just making charts—ideally data cleaning + SQL + analysis + visualization + a business/research question.

Best free dataset sources

• data.gov — Probably my top choice for realistic U.S. projects. It has hundreds of thousands of government datasets covering transportation, health, crime, housing, economics, education, environment, and more.Data.gov

• Great project: “What factors are associated with traffic accidents in U.S. cities?”

• Good for: SQL, Excel, Python, Tableau/Power BI.

• ourworldindata.org — Excellent for economic, demographic, health, environmental, and global-development projects. Many datasets can be downloaded directly as CSV and have detailed source/metadata information.Our World in Data+1

• Great project: “How have life expectancy and GDP changed across countries since 1960?”

• Good for: exploratory analysis, time series, dashboards.

• archive.ics.uci.edu — A long-standing collection with 689 datasets, including business, education, health, retail, and industrial datasets.UCI Machine Learning Repository

• Particularly interesting: Online Retail has about 542K transactions, while Bank Marketing can support customer-conversion analysis.UCI Machine Learning Repository+1

• Good for: SQL, segmentation, customer analytics, predictive analysis.

• kaggle.com — Huge selection and very easy to browse by topic. Look for datasets based on real-world transactions, companies, cities, sports, or economic activity, rather than overly polished tutorial datasets.

• Great for: finding something interesting quickly.

• Tip: Don't simply reproduce the notebook that accompanies a popular Kaggle dataset.

• census.gov — Excellent if you want a serious-looking demographic/economic project involving income, population, housing, employment, education, or geography.

• Great project: “Which demographic characteristics are most associated with housing affordability?”

• bls.gov — Particularly good for employment, wages, occupations, inflation, and labor-market analysis.

• Great project: “How have wages changed across occupations and states?”

• data.worldbank.org — Useful for international economic and development analysis.

• Great project: “Which countries have improved education outcomes most rapidly?”

What I'd choose for a portfolio

If your goal is getting hired as a Data Analyst, I'd rank them:

Source	Portfolio value	Best for
Data.gov	⭐⭐⭐⭐⭐	Real-world business/public-sector analysis
Census	⭐⭐⭐⭐⭐	Demographics, housing, economics
Our World in Data	⭐⭐⭐⭐⭐	Global trends & visualization

One important tip

The dataset matters less than the question you ask of it.

A weak portfolio project is:

“I analyzed the Titanic dataset and made five charts.” A much stronger one is:

“Which factors are associated with customer churn, and which customer segments should a company prioritize for retention?” Then show:

1. Data cleaning
2. SQL queries
3. Exploratory analysis
4. KPI calculations
5. Segmentation
6. Dashboard
7. Key findings
8. Business recommendations

For example, the UCI Online Retail dataset is particularly suitable because it contains hundreds of thousands of actual transactions, giving you enough data to calculate revenue trends, repeat-purchase rates, customer lifetime value, product performance, RFM segments, and cohort retention.UCI Machine Learning Repository

If you tell me whether you're targeting entry-level Data Analyst, Business Analyst, or BI Analyst roles, I can give you 5 specific datasets + a portfolio project question + the SQL/Python/Tableau skills each would demonstrate.

“I analyzed the Titanic dataset and made five charts.” A much stronger one is:

“Which factors are associated with customer churn, and which customer segments should a company prioritize for retention?” Then show:

1. Data cleaning
2. SQL queries
3. Exploratory analysis
4. KPI calculations
5. Segmentation
6. Dashboard
7. Key findings
8. Business recommendations

For example, the UCI Online Retail dataset is particularly suitable because it contains hundreds of thousands of actual transactions, giving you enough data to calculate revenue trends, repeat-purchase rates, customer lifetime value, product performance, RFM segments, and cohort retention.

Absolutely. For a Data Analyst portfolio, I’d prioritize datasets that let you demonstrate more than just charts—ideally cleaning, SQL, exploratory analysis, business questions, and visualization.

Best free dataset sources

• kaggle.com — Probably the easiest place to start. It has hundreds of thousands of public datasets covering sales, marketing, finance, sports, healthcare, entertainment, and more. Many also have notebooks showing how others analyzed them.Kaggle+1
• data.gov — Excellent for real-world analyst projects. You can find U.S. government data on transportation, healthcare, education, demographics, economics, crime, environment, and more. The catalog currently contains over 550,000 datasets.Data.gov+1
• archive.ics.uci.edu — Great for well-documented, relatively manageable datasets. Particularly useful when you want to practice structured analysis without spending hours figuring out what the columns mean.Kaggle
• ourworldindata.org — Excellent for global trends involving health, climate, population, economics, energy, and social issues. Particularly good for time-series and cross-country analysis.Dataquest
• data.worldbank.org — Strong choice for economic and development projects: GDP, poverty, employment, education, population, health, etc.SIIT
• datasetsearch.research.google.com — Useful when you have a specific topic in mind and want to search across many different repositories rather than browsing one site.Kaggle
• data.fivethirtyeight.com — Particularly interesting for portfolio projects because the datasets often connect naturally to a question or story, making it easier to demonstrate analytical thinking.Kaggle

What I'd choose for a portfolio

Don't simply pick the biggest or cleanest dataset. A stronger portfolio project usually has a business or real-world question behind it.

For example:

A particularly good strategy

I'd actually use Data.gov or another government/open-data portal for at least one project rather than making your entire portfolio out of Kaggle datasets. It gives you an opportunity to work with data that is less curated and potentially messier—closer to what analysts encounter in real jobs. Data.gov specifically provides machine-readable government datasets and metadata.Data.gov

For each project, try to show:

1. Business question
2. Data cleaning
3. SQL analysis
4. Exploratory analysis
5. Dashboard in Power BI/Tableau
6. Key findings and recommendations
7. README explaining your methodology

That combination will generally showcase your analyst skills much better than simply downloading a famous Kaggle dataset and making a few charts.Dataquest

If you're building a portfolio specifically to get your first Data Analyst job, I can also give you 10 specific dataset + project ideas, ranked from beginner to impressive, with the SQL/Excel/Power BI skills each one demonstrates.

Absolutely. For a Data Analyst portfolio, I’d prioritize datasets that are real, reasonably messy, and let you answer a meaningful business question—not just datasets that are easy to visualize.

Best places to look

• kaggle.com — Probably the easiest starting point. Huge variety across sales, finance, healthcare, sports, marketing, etc. Many datasets also have notebooks you can use for inspiration.Dataquest
• data.gov — Excellent for authentic U.S. government data: transportation, health, education, economics, crime, demographics, and more. Great for demonstrating that you can work with real-world data.Dataquest
• datasetsearch.research.google.com — More of a search engine than a repository. It searches datasets hosted across thousands of repositories, making it particularly useful when you have a specific topic in mind.Dataset Search
• archive.ics.uci.edu — A well-established collection of datasets. Better suited to structured analytical/ML projects than business storytelling, but still useful.Kaggle
• ourworldindata.org — Excellent for global trends involving health, population, climate, energy, economics, and development. Particularly good for time-series and comparative analysis.Dataquest
• data.worldbank.org — Great for economic and demographic analysis across countries and over time. Google also highlights World Development Indicators as a major public-data source.Google Help
• opendata.cityofnewyork.us — A particularly good source of realistic city-level data, including transportation, inspections, schools, and other public services.Dataquest
• data.fivethirtyeight.com — Useful if you want projects involving sports, politics, surveys, or social trends, with datasets connected to published analyses.Dataquest

What I'd choose for a portfolio

Rather than downloading the first interesting CSV you see, look for a dataset with:

1. A clear question — e.g. "What factors are driving customer churn?"
2. Multiple variables — categorical + numerical + dates are ideal.
3. Some messiness — missing values, duplicates, inconsistent categories, etc.
4. Enough rows to demonstrate SQL/Python/Excel skills.
5. A real-world decision behind it — something a manager could plausibly ask.
6. A source you can cite and explain.

For example:

Project idea	Good data source
U.S. housing market analysis	Data.gov
Customer churn analysis	Kaggle
Global life expectancy trends	World Bank / Our World in Data
NYC transportation analysis	NYC Open Data

One important tip: avoid making your portfolio entirely from overused tutorial datasets such as Titanic or Iris. Current portfolio guidance specifically recommends choosing less-common datasets and framing them around a genuine business question.Dataquest

If you're trying to get hired as a Data Analyst, I'd actually recommend building 3 projects: one SQL-heavy, one Excel/Power BI/Tableau dashboard, and one Python/EDA project, using datasets from these sources.

Absolutely. For a Data Analyst portfolio, I’d prioritize datasets that let you demonstrate more than just charts—ideally you can show data cleaning, SQL, exploratory analysis, statistical reasoning, visualization, and business recommendations.

Best free dataset sources

• data.gov — The U.S. government's open-data portal, with hundreds of thousands of datasets. Great for transportation, health, economics, demographics, crime, environment, and government operations.Data.gov+1
• archive.ics.uci.edu — Excellent for structured/tabular projects. It currently hosts hundreds of datasets, including Online Retail, Bank Marketing, Adult/Census Income, Air Quality, and Student Performance.UCI Machine Learning Repository+1
• ourworldindata.org — Particularly good for global trends involving population, economics, health, energy, climate, and development. Its charts provide downloadable data and metadata.Our World in Data
• Kaggle Datasets — Huge variety and particularly useful for finding realistic, messy datasets and seeing how other analysts approach them.
• Google Dataset Search — Useful when you have a very specific topic in mind and want to discover datasets hosted by universities, governments, research organizations, etc.
• City/state open-data portals — Often underrated for portfolios. Local transportation, housing, permits, 311 calls, restaurant inspections, property values, and crime data can make for very practical projects.

Dataset ideas I'd recommend for a portfolio

One particularly good starting point

The UCI Online Retail dataset is a strong portfolio choice because it contains roughly 542,000 transactions from a UK online retailer. That gives you enough complexity to demonstrate cleaning, SQL aggregation, customer segmentation, revenue analysis, product performance, and time-based trends.UCI Machine Learning Repository

A project could answer questions like:

Which customers and products drive revenue, how does purchasing behavior change over time, and where should the company focus its retention efforts? That sounds much more like an actual Data Analyst business problem than simply "I analyzed a dataset."

Portfolio tip: Don't choose a dataset merely because it's interesting. Choose one where you can formulate a clear business question and finish with 3–5 actionable recommendations. That's usually much more impressive to an employer than a complicated model.