Yesterday's UK 🚦 how were my projections & what does the data used for decisions look like?

Including CTA and removing silly places we ended up with 14 on the 'practical' Green list

Of the 14 I correctly got guessed 12, but of the 24 I predicted only 12 were Green

1/
At best a mixed result, but lets see if we can learn from it to do better in 3 weeks time

The Gov published some data here - https://www.gov.uk/government/publications/data-informing-international-travel-traffic-light-risk-assessments/risk-assessment-methodology-data

Showing their decisions but rather unhelpfully it doesn't show the countries that didn't make Green .. only those moved.

2/
A few points are obvious

- Heavy reliance is placed on post travel Test & Trace data .. it's positive rate and variant sequencing outcomes.

- Public metrics match my guesses, though not sure sequences uploaded to GISAID is that useful for genomic sequencing ability proxy?

3/
Overall hard to make sense of Test & Trace measure given values of 2 or less are redacted as *

Of around 2000 travellers in last month tested coming back from the new Green countries nly 3 positives are reported definitely ( and perhaps a few more redacted )

4/
Equally though from the Red Zone countries not exactly that many testing positive (about 52 out of 2341 ) and nothing above 'redacted' level showing up as indicating variant presence.

Without the counter factual of the countries that didn't move can't say much more

5/
Other than Test&Trace metrics they are using

- Case incidence
- Test Rates
- Test Positivity
- Vaccine Doses

All are in my 'Traffic light score'

Also # of sequences uploaded to GISAID but given they approved Falklands with 0 and Brunei with 10 guess this isn't important.

6/
So in summary

The Variant stuff & returned traveller metrics will continue to be a 'wildcard' as we can't get that data

Other stuff pretty easy to predict

Lack of transparency on coutnries that didn't move makes analysis hard as you can't work out thresholds.

7/
Back to the projections and where I went wrong and where it looks like they made odd decisions

Malta - Makes no sense to have left Malta off 🟩 - high vaccine coverage, low rates good testing - can't explain this oversight. Test & Trace data for Malta won't fill the gap

8/
Finland & Norway - Perhaps lower vaccination rates hurt them, and slightly higher incidence in Norway, but still seems very harsh ( and on the vaccination rate they approved Brunei and Aus/NZ where almost nobody's been 💉) .. Again struggle with their decision here to deny 🟩

9/
Further afield it's a bit more random (and irrelevant for UK travellers) but

Saudi Arabia should have had a good chance, not sure why it was denied

Fiji, South Korea, Taiwan.. hard to see how Brunei, Singapore & Oz/NZ got 🟩 but these didn't, little difference in stats.

10/
My remaining🟩 relied on them being less strict and I think are explainable by the bar being set too high at the start

That's Antigua, Barbados & St. Lucia

Otherwise I missed selecting Brunei & Faroes for 🟩 as I assumed 🟨 (I probably didn't think about them too much)

11/
As regards the 🟥additions (& non removals) it's hard from the data to see all the logic but thoughts here-

Nepal - Hard to argue with this given situation there currently, had I thought about it more I probably would have suggested 🟥too

12/
Maldives -Thought they would escape as they are small in terms of risk size. Test & Trace data in the report doesn't show any issue and they have good vaccine coverage, albeit high case rate now.

Victims of being near India on the map I guess, but then why not Sri Lanka ?

13/
Turkey - I flagged this as a risk a while back https://twitter.com/EvilDoctorK/status/1382619383785271299?s=20 but I thought that the risk had subsided now as their figures went down and as they weren't added at the same time as India then they wouldn't be.

14/
UAE & Seychelles

I thought they both had a shot at escaping to 🟨. As they didn't change tier we have no data to look at unfortunately.

I suspect UAE it's still the 'hubs' problem driving it unfortunately. Not clear short term how that gets fixed.

15/
Also to mention

- No Island Rule ( no surprise in my book as I've been saying ) .. may come yet, but no sign of it yet.

Update to the ratings in 3 weeks ( from yestedray or from 17th May?) let's see if we can do better next time!

I'll do the normal stats thread later!

/ends
You can follow @EvilDoctorK.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: