The penumbra of open source: projects outside of centralized platforms are longer maintained, more academic and more collaborative

EPJ Data Science

Table 2 Comparison of Penumbra and GitHub datasets

Statistic	Fig.	Penumbra			GitHub			KS 2-Sample
Statistic	Fig.	Mean	Median	CI	Mean	Median	CI	KS S	KS P
Files	2(c)	244.47	12	[1,859]	156.07	9	[1,264]	0.07	<0.001
Committers	2(b)	2.39	1	[1,6]	2.08	1	[1,3]	0.17	<0.001
Message Lengths	2(f)	29.24	20.80	[7.00,67.33]	24.23	17.60	[7.42,56.00]	0.13	<0.001
Editor Density	2(d)	1.12	1.00	[1.00,1.60]	1.05	1.00	[1.00,1.30]	0.20	<0.001
Burstiness	3(d)	4.86	2.88	[0.50,14.51]	3.68	2.15	[0.17,11.24]	0.13	<0.001
Commits	2(a)	67.12	8	[1,194]	25.27	4	[1,57]	0.20	<0.001
Branches	2(e)	1.74	1	[1,4]	1.67	1	[1,5]	0.03	<0.001
Age (hours)	3(a)	5528	883	[0.1,25556]	2669	73	[0.03,16194]	0.26	<0.001
Age / Commits	3(b)	283	39	[0.02,1261]	193	9	[0.01,944]	0.19	<0.001
Avg. Interevent	3(c)	375	43	[0.05,1547]	257	11	[0.02,1130]	0.19	<0.001
Team Size	3(e)	1.71	1.00	[1.00,3.92]	1.42	1.00	[1.00,2.67]	0.17	<0.001

Mean, median, and 5^th and 95^th percentile values from the Penumbra and GitHub samples for each statistic. KS S and KS P represent the Kolmogorov-Smirnov two-sample statistic, and its corresponding p-value.