Data Resources

• Simplemaps World Cities Database is a narrowly focused resource that provides high quality data regarding cities, countries and populations. The free dataset they provide may be useful for many projects that involve analysis of geographically related data.

• Our World in Data is both a journal and a data repository. It is focussed on geographic distribution of human related attributes, such as life expectancy, health and economics. It provides a very good way to investigate data analysis and visualisation since all the charts in the publications enable you to download the data from which they were created.

• The UK Met Office website hosts several data sources relating to weather and climate. Data and information on how to use it can be obtained via their datapoint service web page.

• US Geological Survey (USGS) provides many geological and geographic datasets, including live stream data of several kinds. (Detailed geographic data relating to Great Britain is available from the UK's Ordnace Survey via their open data resource, but is a little more difficult to access and utilise than the USGS data.)

• The US National Aeronautics and Space Administration (NASA) provides a data resource hub that enables access to a huge variety of scientific data. This of course includes data relating to astronomy and space science, but also covers data from all fields of physical science. The site provides a visualisation of the different kinds of data that are available.

• Kaggle is currently the largest and most widely used resource providing datasets for data science. It is not just a dataset reporsitory but also supports a community within which data scientists can share and explore each others data. What is more it provides a cloud-based Jupyter notebooks environment and computational resources so that researchers can implement data analysis projects within Kaggle itself. It also facilitates browsing and discover of datasets via its search interface. Kaggle is very much oriented towards the Machine Learning approaches to data analysis and provides many features that support this. You will need to make an account on Kaggle in order to be able to make full use of its facilities, but that is not necessary for just searching or browsing the datasets.

• data.world is similar to Kaggle in supporting a data sharing community rather than just being a repository. It differs from Kaggle in that it is not particularly oriented to ML. It is geared towards more established ways of accessing and analysing and provides facilities for accessing remotely stored data by means of queries formulated in SQL or SPARQL.

• For those interested in using machine learning techniques for data analysis in Python, the Scikit-learn package is an excellent framework within which to conduct such reasearch. It also provides a detailed user guide and tutorials, as well as access to many datasets.

• The prestigeous science journal Nature curates a web page of recommended data repositories, which lists and describes a large number of repositories, covering a very wide range of disciplines. Whereas, the links suggested above were somewhat biased towards geographic and physical sciencs, Nature's list of recommended repositories also contains many links to repositories of biological, medical, and social sciences data.