some short demo content; patchwork fixes to problems found for demo-ing, change later

Author: O957

assets/paste_bin.txt

@@ -38,8 +38,39 @@ REMOVE plot and comparison functions for now
 
 
 
+seed = 54321
+seed = 1
+seed = 213123421 STUCK!
 
 
+213123421 works with
+
+find_heuristic_step_size: True
+
+seed = 2412
+seed = 32452342
+seed = 1234
+
+All seem to work
+
+Now AL on 2024-03-30
+
+seed = 54321
+seed = 12312421
+
+
+            # save to folder for jurisdiction,
+
+            # idata = az.from_numpyro(model.mcmc)
+            # diagnostic_stats_summary = az.summary(
+            #     idata.posterior,
+            #     kind="diagnostics",
+            # )
+            # print(diagnostic_stats_summary.loc["negbinom_rv"])
+
+
+DHM: uniform is good for checking if you are missing constraint that should be present
+
 
 seeding ("initialization" in MSR lingo):
 

config/params_2024-01-20_historical.toml

@@ -22,15 +22,15 @@ reciprocal_dispersion_prior_scale = 30
 reference_day_of_week = 4
 rt_intercept_prior_mode = -0.4054651
 rt_intercept_prior_scale = 0.3
-seed = 54321
+seed = 1234
 susceptible_fraction_prior_mode = 0.9
 susceptible_fraction_prior_scale = 0.1
 weekly_rw_prior_scale = 0.25
 first_fitting_date = "2023-09-15"
 
 
-adapt_delta = 0.85
-max_treedepth = 9
+adapt_delta = 0.80
+max_treedepth = 10
 n_chains = 1
 n_warmup = 1000
 n_iter = 1000

config/params_2024-03-30_historical.toml

@@ -0,0 +1,41 @@
+day_of_week_effect_prior_modes = [0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
+day_of_week_effect_prior_scales = [0.25, 0.25, 0.25, 0.25, 0.25, 0.25]
+holiday_eff_prior_mode = -0.2
+holiday_eff_prior_scale = 0.2
+post_holiday_eff_prior_mode = 0.0
+post_holiday_eff_prior_scale = 0.2
+recency_eff_prior_mode = 0.0
+recency_eff_prior_scale = 0.1
+recency_effect_length = 0
+
+generation_time_dist = [0.233, 0.359, 0.198, 0.103, 0.053, 0.027, 0.014, 0.007, 0.003, 0.002, 0.001]
+ihr_intercept_prior_mode = -6.5
+ihr_intercept_prior_scale = 0.5
+inf_model_prior_infections_per_capita = 0.0001
+inf_model_seed_days = 11
+inf_to_hosp_dist = [0.05, 0.1, 0.175, 0.225, 0.175, 0.125, 0.075, 0.05, 0.025]
+max_rt = 3.0
+n_pre_observation_days = 14
+non_obs_effect_prior_mode = -50
+non_obs_effect_prior_scale = 0.01
+reciprocal_dispersion_prior_mode = 10
+reciprocal_dispersion_prior_scale = 5
+reference_day_of_week = 4
+rt_intercept_prior_mode = -0.4054651
+rt_intercept_prior_scale = 0.3
+seed = 12312421
+susceptible_fraction_prior_mode = 0.9
+susceptible_fraction_prior_scale = 0.1
+weekly_rw_prior_scale = 0.25
+first_fitting_date = "2023-09-15"
+
+
+
+adapt_delta = 0.80
+max_treedepth = 10
+n_chains = 1
+n_warmup = 1000
+n_iter = 1000
+
+
+exclude = ["HI"]

config/params_2024-04-13_historical.toml

@@ -0,0 +1,37 @@
+day_of_week_effect_prior_modes = [0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
+day_of_week_effect_prior_scales = [0.25, 0.25, 0.25, 0.25, 0.25, 0.25]
+holiday_eff_prior_mode = -0.2
+holiday_eff_prior_scale = 0.2
+post_holiday_eff_prior_mode = 0.0
+post_holiday_eff_prior_scale = 0.2
+recency_eff_prior_mode = 0.0
+recency_eff_prior_scale = 0.1
+recency_effect_length = 0
+
+generation_time_dist = [0.233, 0.359, 0.198, 0.103, 0.053, 0.027, 0.014, 0.007, 0.003, 0.002, 0.001]
+ihr_intercept_prior_mode = -6.5
+ihr_intercept_prior_scale = 0.5
+inf_model_prior_infections_per_capita = 0.0001
+inf_model_seed_days = 11
+inf_to_hosp_dist = [0.05, 0.1, 0.175, 0.225, 0.175, 0.125, 0.075, 0.05, 0.025]
+max_rt = 3.0
+n_pre_observation_days = 14
+non_obs_effect_prior_mode = -50
+non_obs_effect_prior_scale = 0.01
+reciprocal_dispersion_prior_mode = 10
+reciprocal_dispersion_prior_scale = 5
+reference_day_of_week = 4
+rt_intercept_prior_mode = -0.4054651
+rt_intercept_prior_scale = 0.3
+seed = 54321
+susceptible_fraction_prior_mode = 0.9
+susceptible_fraction_prior_scale = 0.1
+weekly_rw_prior_scale = 0.25
+first_fitting_date = "2023-09-15"
+
+
+adapt_delta = 0.8
+max_treedepth = 10
+n_chains = 1
+n_warmup = 1000
+n_iter = 1000

data/2024-04-13/2024-04-13_config.toml

@@ -0,0 +1,37 @@
+day_of_week_effect_prior_modes = [0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
+day_of_week_effect_prior_scales = [0.25, 0.25, 0.25, 0.25, 0.25, 0.25]
+holiday_eff_prior_mode = -0.2
+holiday_eff_prior_scale = 0.2
+post_holiday_eff_prior_mode = 0.0
+post_holiday_eff_prior_scale = 0.2
+recency_eff_prior_mode = 0.0
+recency_eff_prior_scale = 0.1
+recency_effect_length = 0
+
+generation_time_dist = [0.233, 0.359, 0.198, 0.103, 0.053, 0.027, 0.014, 0.007, 0.003, 0.002, 0.001]
+ihr_intercept_prior_mode = -6.5
+ihr_intercept_prior_scale = 0.5
+inf_model_prior_infections_per_capita = 0.0001
+inf_model_seed_days = 11
+inf_to_hosp_dist = [0.05, 0.1, 0.175, 0.225, 0.175, 0.125, 0.075, 0.05, 0.025]
+max_rt = 3.0
+n_pre_observation_days = 14
+non_obs_effect_prior_mode = -50
+non_obs_effect_prior_scale = 0.01
+reciprocal_dispersion_prior_mode = 10
+reciprocal_dispersion_prior_scale = 5
+reference_day_of_week = 4
+rt_intercept_prior_mode = -0.4054651
+rt_intercept_prior_scale = 0.3
+seed = 54321
+susceptible_fraction_prior_mode = 0.9
+susceptible_fraction_prior_scale = 0.1
+weekly_rw_prior_scale = 0.25
+first_fitting_date = "2023-09-15"
+
+
+adapt_delta = 0.9
+max_treedepth = 12
+n_chains = 1
+n_warmup = 1002
+n_iter = 2000