May I suggest downplaying the downloads stat? I'm assuming you get this from summing up the downloads on pypi but it only sort-of correlates with usage. Many people with pip/virtualenv will download the same packages over and over as they redeploy. It also tends to favor old projects.

I think a better metric for popularity would be a blend of activity in the source code repos, activity on mailing lists, wikis, blog postings etc. I liked djangopackages' use of a "I use this" metric because it helped me make comparisons across github and bitbucket. Much harder to gather but much more useful.

Edit: Also another issue with downloads can be seen with something like the Requests library. It has 40k downloads. Is that a lot? What do you compare it to? For me the important questions are: are people moving to it or away from it? Is it active or not? How popular is it in comparison to other libraries that do similar things?

Good luck on this project. It seems like a good start.