Peter Cayton, the Stats Guy Profile picture
Statistician. True Neutral. 147. Public Twitter of Assoc Prof Peter Julian Cayton, PhD, @UPDStat @Official_UPD. Tweeted views are my own. He/They. 🇵🇭 75577819
May 3, 2022 22 tweets 5 min read
Ensemble approaches, also known as combination methods, exist in statistics and data science to try & make sense of predictions coming from different models by combining them through some scheme.

In some instances, combining models by averaging predictions can be accurate! Ensemble predictions tend to perform well as a means to settle highly varying estimates.

When one model prediction is too high,

And another model prediction is too low,

Sometimes the average of the two may be just right.
May 2, 2022 7 tweets 2 min read
Daily Change(+/-%) & Crude Days-to-Double

> Conf = 188, 253 days
>> 1 case purged, reflecting an increase of 187 (+0.0051%) in total cases
> Recov = 838 (+0.0232%), 245 days
> Deaths = 36 (+0.0215%), 261 days
> Net daily change in active = -664 Source: doh.gov.ph/covid19tracker

Personally-Archived Data: docs.google.com/spreadsheets/d…

Highest recovery rate ever.

Highest case fatality rate since Jan 16, 2022.

Lowest no. of active cases since April 19, 2020.
May 2, 2022 4 tweets 2 min read
Let's discuss this one:

Question of Age Representation

Here below is the estimated voting population with percentage breakdown by age, and Pulse Asia's percentage breakdown.

Pop'n nos. used are 2022 projections.

I estimated the 18-19 age group as 40% of the 15-19 age group. Yes, Pulse Asia seemed to have underrepresented the 18-34 group and overrepresented the 45 and older. Kinda nailed it in the 35-44 range.

What if we add the estimated social media internet users in the Philippines?
May 2, 2022 5 tweets 2 min read
The website of the April 2022 survey by Pulse Asia.

pulseasia.ph/april-2022-nat… And additional breakdowns

pulseasia.ph/pb-april-2022-…
Apr 11, 2021 5 tweets 1 min read
"In a rare admission of the weakness of Chinese coronavirus vaccines, the country’s top disease control official says their effectiveness is low and the government is considering mixing them to give them a boost."

apnews.com/article/beijin… '... Chinese vaccines “don’t have very high protection rates,” said the director of the China Centers for Disease Control, Gao Fu, at a conference Saturday in the southwestern city of Chengdu. ...'
Apr 10, 2021 4 tweets 2 min read
Note:
> 19 case purged
> 56 recovs purged/changed status
> 1 death purged/changed status

Daily % Change & Crude Days-to-Double
Conf= 1.5056%, 134 days
Recov= 0.0829%, 168 days
Deaths= 1.5427%, 157 days

Net increase in active = 11,894

Data: docs.google.com/spreadsheets/d… Highest no. of active cases ever.

Second highest no. of new cases reported ever.

Lowest recovery rate since Sept 19, 2020.

Recovery is the slowest of the three counts.

The end of the pandemic is far from sight.
Apr 10, 2021 5 tweets 2 min read
I am slightly aggravated that a private firm is on a broadsheet media saying "we only need 35 million vaccinated for herd immunity".

The gall!

So, let's tickle this devilish idea with some simple mathematical equation

H = (1 - 1/R)/E

academic.oup.com/cid/article-pd… H = herd immunity threshold thru vaccination
R = reproduction number
E = vaccine efficacy

Suggested H this truth-misrepresenting private firm is 35 million or 35/110 ≈ 31.8182%.

Currently, the available vaccine has 50.4% efficacy based on FDA's EUA

fda.gov.ph/wp-content/upl…
Apr 10, 2021 5 tweets 3 min read
From covid19.psphp.org by @L4H_COVID19

PHI Rt has dropped below 1 for the first time, but there are still a lot of new cases. Why is that?

Well, compare this week to last week.
Apr 2-9: new cases bet. 6k-12k/day
Mar 27-Apr 1: new cases 6k-15k/day ImageImageImageImage It seems that for the past 2 weeks, we are settling in the same range of new cases/day, as if every 1 case of COVID last week, they tend to infect 1 case of COVID for this week.

Thus, Rt has moved closer to 1, and yesterday, slightly below it.
Apr 9, 2021 4 tweets 2 min read
Our Philippine COVID-19 Active Cases Interactive Dot Maps are now at endcov.ph/map, as a data layer named "Active Cases (Dot Density) Philippines", data as of April 4, 2021. ImageImageImageImage Thanks to the mappers, developers, researchers, and volunteers facilitated by Jake Mendoza, Feye Andal, UP Resilience Institute, and UP COVID-19 Pandemic Response Team!
Apr 9, 2021 4 tweets 2 min read
Note:
> 37 cases purged
> 231 recovs purged/changed status

Daily % Change & Crude Days-to-Double
Conf= 1.4713%, 138 days
Recov= 0.1105%, 167 days
Deaths= 2.8401%, 159 days

Net increase in active = 11,072

Data: docs.google.com/spreadsheets/d… Highest no. of active cases ever.

Highest number of reported deaths ever.

3rd highest no. of new cases. Previously, the top 1 and 2 were in April 2 and 3, respectively

The end of the pandemic is far from sight.
Apr 9, 2021 5 tweets 1 min read
Ay tanga. Anong bago sa ECQ?

Anong improvement ang nagawa?

Meron bang bagong contact tracing system at protocol?

Napalakas na ba ang testing upang umabot sa 180k indivs/day?

May maayos na bang plano sa supply at administration ng bakuna?

May maayos na bang sistema sa ayuda?
Apr 5, 2021 5 tweets 2 min read
Note:
> 8 cases purged
> 8 recovs purged/changed status

Daily % Change & Crude Days-to-Double
Conf= 1.2689%, 145 days
Recov= 0.0161%, 163 days
Deaths= 0.0379%, 167 days

Net increase in active = 8,200

Data: docs.google.com/spreadsheets/d… I am seeing some signs of slowing, but not yet definitive.

Would 8k-10k cases be our new typical number of cases?

The end of the pandemic is far from sight.
Apr 5, 2021 4 tweets 1 min read
(An Email to my Students, April 5, 2021)

Hi, dear student,

I hope you are doing well and staying safe.

This week has not been kind to fragile hearts and I understand what you are feeling as I feel it, too. We are not in a good place and our thoughts may be clouded with the situation we all are in this country.

I want you to know that you are loved and someone thinks of your safety and health.
Apr 5, 2021 5 tweets 2 min read
Average Delays of DOH Data, as of April 4, 2021:

Onset of Symptoms to Specimen Collection = 4.1765 days
Specimen Collection to Result Release = 2.7296 days
Result Release to Official Confirmation = 6.3509 days
Sum of Three = 13.2570 days Recovery to DOH Report of Recovery = 11.6584 days
(Note: only 1/3 of recoveries have date of recovery)
Death to DOH Report of Death = 40.2771 days

Data Source: docs.google.com/spreadsheets/d…
Apr 5, 2021 4 tweets 1 min read
Apr 3 Data As of Apr 4 DOH data drop:

Confirmed = 784,023
Individuals Tested Positive = 870,749

Gap = 86,726

The gap is large. There gap trend is crawling on incline. Now holding the 86k-97k range. Biased Estimate of Validation Rate, Apr 4 = 12.7159%
Biased Estimate of % Confirmed, Apr 3 = 90.0401%

Validation rate has increased, but cases still grow. % Confirmed touches the 90% range.
Apr 5, 2021 18 tweets 4 min read
Compendium of Philippine COVID-19 Statistics
As of April 4, 2021
Volume I: Island Groups, Regions, and Provinces

drive.google.com/file/d/1HnAXLU… PROTEST OF PETER CAYTON:

BRING BACK #DATEREPREM, @DOHgovph!

IT IS NOT REDUNDANT!

SEE FOR YOURSELF WHAT SUCH VARIABLE CAN DO TO HELP THE PEOPLE!

BEST IF YOU ARE DOING THIS, NOT US.

YOU ARE THE AGENCY FOR THIS!
Apr 5, 2021 15 tweets 3 min read
Regional Statistics on COVID-19 Testing
Most Recent Data as of April 3, 2020

drive.google.com/file/d/1vXQfr3… PH Daily Positive (+) Rate: 23.1298%

PH daily (+) rate is in an uptrend since trough in Feb 11, 2021 at 5.2882%. Expanded national testing is needed. The 7-day average positives hitting 9651.71. Testing should be >= 200k heads/day.
Apr 5, 2021 5 tweets 1 min read
Statistics on COVID-19 Dedicated Hospital Capacity,
Most Recent Data as of April 3, 2021
Volume I: Regions and Provinces

drive.google.com/file/d/1HTrPNS… Provinces with High ICU Bed Utilization Rate (>=70%)
[1] "BATAAN"
[2] "BENGUET"
[3] "BULACAN"
[4] "ISABELA"
[5] "LAGUNA"
[6] "METRO MANILA"
[7] "MOUNTAIN PROVINCE" (100%)
[8] "NUEVA VIZCAYA"
[9] "PAMPANGA"
[10] "QUEZON"
[11] "RIZAL" (100%)
[12] "TARLAC"
Apr 4, 2021 4 tweets 1 min read
Deaths have the wildest delays to date of report of death (i.e., when they get reported by DOH on official counts), with an average of 40 days.

Recovery have delays of about 11.72 days from recovery to date of report of recovery, but this is with the 1/3 of cases with dates. Data Source: docs.google.com/spreadsheets/d…
Apr 4, 2021 4 tweets 1 min read
Note:
> 20 case purged
> 10 recovs purged/changed status
> 0 death purged/changed status

Daily % Change & Crude Days-to-Double
Conf= 1.4040%, 147 days
Recov= 6.8102%, 162 days
Deaths= 0.0149%, 166 days

Net decrease in active = 30,189

Data: docs.google.com/spreadsheets/d… The most ridiculous increase in recoveries, ever. The miracle of San Francisco Duque for Easter.

Have mercy on the Philippines.

Lowest recovery rate since Sept 19, 2020.

The end of the pandemic is far from sight.
Apr 3, 2021 9 tweets 1 min read
Ito dapat ang hanapin natin:

1) Contact Tracing
Bagong sistema?
Up to 3rd degree contacts?
Paano papasok ang partisipasyon ng komunidad?
Ano pa ang kulang para mapalakas?
Baseline metrics na batayan ng kalidad? 2) Mass Testing
Libre o heavily subsidized?
Gaano karaming staff ang iha-hire sa publicly-owned labs?
Ilang labs ang aimed na mabuksan?3
Paano popondohan ang kailangang equipment at materials?
Ano ang target maintaining testing/day per region?