Merge pull request #21 from epiforecasts/feature-make-data-updating-m…

…ore-robust

Feature: make data updating more robust

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: omicron.sgtf.forcast
 Title: Real-time estimation of the time-varying transmission advantage of Omicron in England using S-Gene Target Status as a Proxy
-Version: 1.1.0
+Version: 1.2.0
 Authors@R: 
     c(person(given = "Sam Abbott",
            role = c("aut", "cre"),

NEWS.md

@@ -1,3 +1,9 @@
+# 2021-12-28
+
+- Updated data and results archiving to be by date of access rather than by the maximum date available in the data. This has the download of making an update occur every day the pipeline runs regardless of if the data has been updated but the upside of protecting archived data from retrospective updates.
+
+- Updates dating of all archived data and results to be by date of access rather than by maximum available date in the data.
+
 # 2021-12-26
 
 - Automation has been improved so that new data is checked for every two hours. A new model run is only started if not result exist for the current data snapshot or if new SGTF data is detected.

R/munge-data.R

@@ -97,6 +97,7 @@ cumulative_percentage <- function(cases, pop) {
            c_q5 = cumsum(q5) / population,
            c_q95 = cumsum(q95) / population,
            c_q20 = cumsum(q20) / population,
-           c_q80 = cumsum(q80) / population)
+           c_q80 = cumsum(q80) / population) %>%
+    ungroup()
   return(cases_pop)
 }

R/plot-summary.R

@@ -13,12 +13,18 @@ plot_summary <- function(sgtf_posterior, cum_per, bias_posterior,
 
   # omicron transmission advantage
   advantage <- summary(sgtf_posterior, type = "voc_advantage") %>%
-    mutate(value_type = "Omicron: current transmission advantage")
+    mutate(value_type = "Omicron: current transmission advantage") %>%
+    group_by(region) %>%
+    slice_max(date) %>%
+    ungroup()
 
   # growth
   growth <- summary(sgtf_posterior, type = "growth") %>%
     filter(type == "Omicron") %>%
-    mutate(value_type = "Omicron: current growth rate")
+    mutate(value_type = "Omicron: current growth rate") %>%
+    group_by(region) %>%
+    slice_max(date) %>%
+    ungroup()
 
   # cumulative pop
   cumpercent <- cum_per %>%
@@ -28,11 +34,17 @@ plot_summary <- function(sgtf_posterior, cum_per, bias_posterior,
     mutate(value_type = "Omicron: cumulative % population",
            region = as.character(region),
            type = "Omicron",
-           across(starts_with("q"), ~ . * 100))
+           across(starts_with("q"), ~ . * 100)) %>%
+    group_by(region) %>%
+    slice_max(date) %>%
+    ungroup()
 
   # bias
   sampling <- summary(bias_posterior, type = "voc_advantage") %>%
-    mutate(value_type = "Current bias: any-SGT vs no-SGT result")
+    mutate(value_type = "Current bias: any-SGT vs no-SGT result") %>%
+    group_by(region) %>%
+    slice_max(date) %>%
+    ungroup()
 
   # join --------------------------------------------------------------------
   region_factor <- as.character(sgtf_posterior$region)
@@ -46,7 +58,7 @@ plot_summary <- function(sgtf_posterior, cum_per, bias_posterior,
 
   summary_data <- bind_rows(advantage, growth, cumpercent, sampling) %>%
     filter(
-        date == date_forecast_start & type %in% c("Omicron", "SGT-result")
+        type %in% c("Omicron", "SGT-result")
     ) %>%
     mutate(region = recode_factor(region, !!!region_factor),
            ref_line = as.numeric(recode(value_type, !!!reference_lines)))