jan 19

Jan 12

5% of pilot covariance ~ 10% acceptance

10% of pilot covariance ~ 6% acceptance

Mean Comparison:5%/10%

5%/10%

Jan 5

Proposal from posterior variance.

Wave 3:

Wave 2:

ADAPTIVE SCHEME – Running But looks equally shit.

ESS 

 

Jan 2

Wave 2:

Long Run

1

2

3

4

5

7

8

9

9

10

Combined

 

Wave 3:

Long Run

1

2

3

4

5

6

7

8

9

Combined

Dec 19

Fixed Prior Immunity, Psus = proportion of population that is susceptible, where for each particle the number of susceptible is Bino(N,Psus).

Psus = 0.75

Psus = 0.50

Psus = 0.25

Inferring Prior Immunity.

Wave 3:

Combined results

Comparison of fits; With Prior immunity and Without.

With Prior Immunity

Without Prior Immunity

Wave 2:

 

Combined results

Compare Fitting with past attempts at inferring Wave 2 without prior immunity.
https://mphillachlanbubb.wordpress.com/2016/12/08/dec-8/#jp-carousel-1923 

Comparison of fits; With Prior immunity and Without.

With Prior Immunity

Without Prior Immunity

Dec 14

Fitting with and without the Ceramic.

 

Large Phi boundaries

Truncated Ceramic

ATTACK RATES

Dec 12

Increased MaxTime without Ceramic
Increased standard (if unrecorded) MaxTime to 10 days instead of 5. Still shows bi-modality with high Pc values, but maybe less prominent.

Ceramic Individual tests
Each of these tests didn’t have much of an effect on high Pc.

Condition on 2nd Day first observation.

Condition on 3rd Day first observation.

Increase possible MaxTime to 10 days instead of 5.

Within parallel tests
Increasing MaxTime on all ships including Ceramic.

Increasing MaxTime on all ships and using Condition on 2nd day for Ceramic

Increased PHI tests

Dec 8

DATASETS

Wave 3 without Ceramic
Still displays high Pc values, but appears to be multi-modal and jumps in and out.

For comparison Wave 3 with Ceramic.
Stays in high Pc values, never jumps out.

 

Wave 2 Extended Run

Dec 5

New Inference Scheme:

Wave 2:
New scheme is working well for wave 2. Able to reasonably identify Phi_S and Phi_A.

Wave 3:
Wave 3 has gone to shit doing all sorts of wacky things. Pc values of 99%, Phi_A values near 2 => Extremely high Beta_A values.

I believe mostly caused by relaxing the assumption that asymptomatic are transmit at a lower rate than symptomatic. The weird values seem to come from the wave 3 ship the Ceramic. The individual fit of the ceramic is given below. Displays the high Pc and Phi_A values.

Compared to the other wave 3 ship individual runs (below), inferring Phi_S or Phi_A from a single outbreak is pretty much impossible (as with the Wave 2 ships too).  Inference seems dominated by the Ceramic?

Dec 1

RHOP in [0,30]

Wave 2:

Wave 3:

RHOP in [0,50];

Simulation study