A dataset is a structured collection of data organized in a specific format. Datasets in engineering can come in various forms, including numerical data, experimental results, measurements, simulations, and more.
Despite the fact that funding agencies now require that research data is discoverable, accessible, and preserved for further use, finding data for re-use and analysis can be time-consuming and complicated.
Use the following sources as a starting point for helping to find sample datasets to practice analytical, manipulation, or visualizing skills.
Deep Blue Data
The University of Michigan Library has developed Deep Blue Data, a repository for sharing and archiving research data that were developed at the University of Michigan.
Using the Browse button on the homepage allows you to explore the 1000+ datasets currently hosted in Deep Blue Data by Creator, Language, or Discipline:
One of the advantages of Deep Blue Data is its location on our campus. Creators can be contacted on campus and consulted on their data, offering clarifications, context, and perhaps the possibility of collaboration.
Dimensions Analytics
Dimensions Analytics has included datasets as a content type since 2020. Data are sourced from Figshare as well as Dryad, Zenodo, Pangaea, NIH, and other repositories. “Datasets” is one of the featured document types on the landing page, and selecting this will highlight the most recent datasets indexed by Dimensions:
Dimensions offers a traditional set of filters for datasets, including Funding Agency, Repository, and Source Title.
ICPSR
Inter-university Consortium for Political and Social Research
Located on the University of Michigan campus, ICPSR is an international consortium of more than 750 academic institutions and research organizations. They host specialized collections of data in education, aging, criminal justice, substance abuse, terrorism, and other fields.
Use the Find Data dropdown menu to select Find Data directly.
On the Results page, the Filters column on the left-hand side of the screen allows you to narrow down Data Formats to SPSS, R, delimited, Stata, etc.
The Resources for Students page (under the Teaching & Learning dropdown) has guides to interpreting SPSS output, how to cite data properly, and webinars and tutorials.
Google Dataset Search
Google launched Dataset Search in 2018. Dataset Search searches websites and data repositories for datasets. Subject coverage is multidisciplinary and includes government, scientific, and commercial datasets. About 45 million datasets have been indexed.
Search is straightforward in Dataset Search, although the search results page is formatted differently than in Google or Google Scholar. Filters are along the top, and results are vertically aligned in a scrollable frame on the left. Datasets pop open on the right when selected from the results list.
Filter options in Google Dataset Search are not as robust as in other databases, but they do allow limiting by Download Format in the toolbar above the results. Options include Tabular, Document, Image, etc, including "croissant" which is a format for machine learning datasets.
The Free filter can be toggled on and off. With Free toggled off, Dataset Search will pull data from inside library subscription databases like Statista and commercial sources like Global Data.
Web of Science — Data Citation Index
Data Citation Index was launched inside Web of Science in 2012. It is a stand-alone product and is not included in the WoS Core Collection. It features research data from a wide range of international data repositories in the sciences, social sciences, and humanities. Data Citation Index indexes more than 14 million datasets.
To access this resource, use the Web of Science dropdown menu to select “Data Citation Index”.
Typical WoS filters like Year, WoS Category, Affiliation, and Subject can be applied after running a search in Data Citation Index. Unique to this index is the filter Data Type, which features a wide array of choices like Meteorological, Nucleotide Sequencing, Molecular Structure, Images, Software, and a hundred more.
The results for a given dataset will include a link to the data if available, plus standard metadata and a link in the sidebar to any WoS-indexed papers citing the data. These links will go to WoS Core Collection records.
Sage Research Methods
Sage Research Methods Datasets launched in 2015. This resource is excellent for learning a new statistical technique, as each dataset comes with an explanation of the analytical approach and an example to demonstrate the method. More than 600 datasets are available.
Datasets are separated into either Quantitative or Qualitative collections, and they include Engineering in their discipline filter. Engineering datasets demonstrate techniques like simple and multiple regression using Stata and SPSS.
You can browse datasets by Discipline, Method [chi-square, linear regression], Quantitative [network, spatial, numeric], and Qualitative [audio, diary, drawings].
With Sage Datasets you can:
- Practice qualitative and quantitative analytical methods with sample data and instructional guides
- Find datasets that demonstrate dozens of methods and represent work in Business, Education, Health, Political Science, Psychology, and Sociology
- Find downloadable data to use in course assignments and exam questions