This site is an attempt to create a visualization of the relationships between the ~10,000 most-starred (as of late February 2016) public repositories on GitHub. These repositories were clustered into an ecosystem hierarchy - each circle represents a cluster, and you can expand each cluster by clicking or tapping it. On the desktop, hovering over a circle shows some of the contents and clicking zooms in. On mobile, tapping once shows the contents above the map and tapping again (or holding) zooms in. Once you've zoomed in, each white dot corresponds to a single repository.
To look up a specific project, enter the name of the repo in the search box, or click on its white dot after you've zoomed in far enough. Once you've selected one, its location in the tree will be outlined in bright yellow, and related repositories will be outlined in purple on the map and listed in the "Repos related to..." section. The white dot representing selected repo will turn yellow, and the dots for related repos will turn purple. Clicking the yellow (selected) dot again, or clicking the X in the select dropdown will clear the selection. Clicking on a repo in the "Repos related to..." section will select that repo.
The hierarchy is created by iteratively applying affinity propagation to the repositories, using a PageRank-inspired formula to calculate the strength of the similarities between projects. In a nutshell, repo A has an affinity for repo B if repo A's contributors also contributed to, or starred, repo B. These affinities are not symmetric - repo A's affinity for repo B is not equal to repo B's affinity for repo A. Two repositories are considered related if repo A's affinity for repo B, and repo B's affinity for repo A, are both above a threshold. More details here, or you can read the code.
I downloaded the data and performed the clustering analysis in python, using PyGithub to access the GitHub API. The site itself is mostly written using D3, with the main interface initially coming from this block.
The results of this algorithm are only one possible way of visualizing the data, and clearly there isn't a strict ecosystem hierarchy (should react-bootstrap be classified with React or with Bootstrap?). Almost all of the repos here are related to projects in different clusters, and there are a lot of ways to try and improve the algorithm. There are 10,000 projects included on this map, so I'm sure there are many that could be put in a better place - if you devise a better algorithm, feel free to submit a pull request! But overall the results seem fairly intuitive to me, and hopefully you can use this site to learn about useful projects that you hadn't known about before and have fun looking up your favorite projects and exploring the tech stacks you work with.
I'm not affiliated with GitHub in any way.