#stats

tpq1980@iviv.hu
isha@pluspora.com

Mortality Analyses - Johns Hopkins Coronavirus Resource Center

For anyone thinking that the US is doing badly with the Covid-19 epidemic and believes fairy tales of how well the UK is doing: The US may be climbing up the statistic poles fast, and some states opening up far too early, but look at the top line of the bar chart "Mortality: Observed case-fatality ratio". And at the "Cases and mortality by country" table.

Mortality in the most affected countries

For the twenty countries currently most affected by COVID-19 worldwide, the bars in the chart below show the number of deaths either per 100 confirmed cases (observed case-fatality ratio) or per 100,000 population (this represents a country’s general population, with both confirmed cases and healthy people). Countries at the top of this figure have the most deaths proportionally to their COVID-19 cases or population, not necessarily the most deaths overall.

https://coronavirus.jhu.edu/data/mortality

#covid19 #covid-19 #CoronaVirus #pandemic #deaths #stats #statistics #MortalityRates #health #UK #US #UKPolitics

dredmorbius@joindiaspora.com

G+ Communities: Significant and Recent Activity

We're continuing to get a sense of G+ Communities and how many of these are significant, active, and have some sense of vitality.

One metric is to look by both size and recent activity.

G+ Public Communities with Recent Activity

Members Week Month
1,000 28,823 39,301
100 63,874 105,166

Members is the reported membership of the community on its Google+ landing page. Comparison is greater than or equals.

Recency is assessed as the sampling time minus the most recent post activity, both expressed as Unix timestamps (seconds from 1970-01-01:00:00:00). Week is activity within 86000 * 7 seconds, Month is activity within 86000 * 31 seconds. Comparison is less than or equals.

From previous analysis, it's likely that only 10-20%, generously, of these communities have substantive interaction between users, as opposed to other activity, including spam or link-drops. We have not investigated this dataset yet to assess this characteristic.

Still, that would leave about 3,000 - 6,000 communities of 1,000+ members with weekly or better activity, and 4,000 - 8,000 with monthly activity, and 6k - 12k and 11k - 22k of 100+ members respectively.

There is little relationship between active engagement, measured by plus-one, comment, and reshare activity, and community size. While smaller communities (especially below 100 members) are far more likely to have no activity or no recent activity, the distribution of significant recent activity is fairly uniform across a very broad range of community sizes. On a per member engagement level, the sweet spot appears to be roughly 100 - 5,000 members, though we need to investigate this further.

Very large G+ communities are frequently not particularly engaged. See the WhatsApp Community (6.8 million members) as an example. Interestingly, this doesn't appear in a G+ Community search for "WhatsApp".

Gory Details

URLs for 8,107,099 Google+ Communities were read, with 3,559 found to be deleted when accessed, and 8,103,540 read successfully yielding data.

There were 7,303,513 public, and 800,382 private communities (9.877% private). Membership and post activity are available only for public communities.

Membership status was open ("Join") for 3,689,488 (45.51%), and closed ("Ask to join") for 4,414,407 (54.45%) communities, status was unavailable for 3,204 (0.04%) communities.

Membership percentiles (preliminary):

Memb. Policy 1% 5% 10% 25% 33% 50% 67% 75% 90% 95% 99%
All 0 1 1 1 1 1 4 8 40 110 859
"Join" 0 1 1 1 1 1 1 2 17 50 380
"Ask" 0 1 1 1 1 3 7 13 58 158 1253

(NB: we want to confirm that the Join and Ask to join stats are correct as previous analysis showed higher membership for open-membership sites, though that sample excluded the very largest G+ communities. It's possible sense is inverted. Spot checks say we're not fouling this one up.)

Source

Web-scraped query of 8,107,099 Google+ Community landing pages between 5 Jan and 6 Jan, 2019.

#googlePlus #communities #stats #rStats #plexodus #gplusRefugees