Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on May 29, 2020

• Latest test data.

  

  Will M. Farr

  committed
  108a03c

• Don't need total counts in model any more.

  

  Will M. Farr

  committed
  aadfbf8

• Moving to negative binomial model with cutoff at phi = 100.

  

  Will M. Farr

  committed
  32a1784

Commits on May 19, 2020

• Sampling using binomial model. Curves suck.

  

  Will M. Farr

  committed
  d2de618

• Moving to binomial model. Key realization: smooth pos and neg timeseries and *then* add.

  

  Will M. Farr

  committed
  f6a47fd

Commits on May 18, 2020

• Update.

  

  Will M. Farr

  committed
  9541f37

Commits on May 8, 2020

• Now show positive daily tests as well as R_t estimates.

  

  Will M. Farr

  committed
  72f8887

• Nightly.

  

  Will M. Farr

  committed
  2caf396

Commits on May 7, 2020

• Nightly---a daily adjustment to the timeseries didn't work out.

  

  Will M. Farr

  committed
  22e2c0a

• WIP, adding daily adjustment factors and want to test on laptop.

  

  Will M. Farr

  committed
  1730e99

Commits on May 5, 2020

• Nightly.

  

  Will M. Farr

  committed
  f5afc55

Commits on May 4, 2020

• Daily.

  

  Will M. Farr

  committed
  9aa9beb

Commits on May 3, 2020

• Re-run on nightly data. Bleak picture now that sigma is more restricted---most states above 1.

  

  Will M. Farr

  committed
  a3bed8c

• Start off the Rt_raw series with exp_cts[1] = L0 instead of k[1].

  

  Will M. Farr

  committed
  147e5ca

• Change scale in log_sum_exp linearization for Rt in likelihood-dominated fit.

  

  Will M. Farr

  committed
  561ae2b

• Account for log_sum_exp floor in Rt likelihood s.d. estimate.

  

  Will M. Farr

  committed
  0ae99fd

• Moving away from neg_binomial, also tighter prior on sigma.

  

  Will M. Farr

  committed
  2e69458

Commits on May 2, 2020

• Moved to negative binomial from Poisson; not much difference.

  

  Will M. Farr

  committed
  50c1a59

Commits on May 1, 2020

• Run nightly.

  

  Will M. Farr

  committed
  43a462f

Commits on Apr 30, 2020

• Back to Poisson because total testing numbers are so error-filled.

  

  Will M. Farr

  committed
  34e9dcf

• Saving binomial model in notebook, but returning to Poisson model because most states total test data is crap.

  

  Will M. Farr

  committed
  ddd1e77

Commits on Apr 28, 2020

• More commentary in notebook.

  

  Will M. Farr

  committed
  34a75a5

• Run on today's data using binomial model, but it's *ugly*.

  

  Will M. Farr

  committed
  6ce3944

• Moving to binomial model.

  

  Will M. Farr

  committed
  2b64198

Commits on Apr 27, 2020

• Sampling finished.

  

  Will M. Farr

  committed
  db910d8

• Ready for run on Rusty---sampling is taking longer than I would like.

  

  Will M. Farr

  committed
  fb96a4c

Commits on Apr 26, 2020

• WIP: trying to get good sampling---problem with samlping in Rt space.

  

  Will M. Farr

  committed
  5b0bde9

Commits on Apr 25, 2020

• Moved back to gaussian smoothing.

  Show description for 7093f6c
  

  Will M. Farr

  committed
  7093f6c

• Changed smoothing timescale to actual numbers based on delay times.

  

  Will M. Farr

  committed
  5a2ee04

• Moved to causal exponential smoothing, based on k-sys#30.

  

  Will M. Farr

  committed
  a756d31

• Merge remote-tracking branch 'upstream/master'

  

  Will M. Farr

  committed
  2f10432

• Major performance improvements via re-parameterization to take account of days where the AR(1) prior dominates and days where the observational likelihood dominates.

  

  Will M. Farr

  committed
  22c277e

Commits on Apr 24, 2020

• Move to JHU data, update results.

  

  Will M. Farr

  committed
  a2dcbdb

• Adding Gaussian Process MCMC (beta)

  

  KS

  committed
  fac27a2

• Better re-try options for fit.

  

  Will M. Farr

  committed
  3b55db9

Pagination

PreviousNext