update report

analysis/report.R

@@ -0,0 +1,30 @@
+library("tidyverse")
+library("here")
+library("arrow")
+library("ggplot2")
+library("data.table")
+
+## create output directories ----
+fs::dir_create(here("analysis"))
+
+#define study start date and study end date
+source(here("analysis", "design", "design.R"))
+args <- commandArgs(trailingOnly = TRUE)
+study_start_date <- as.Date(as.numeric(study_dates[args[2]]), format = "%Y-%m-%d", origin = "1970-01-01")
+study_end_date <- as.Date(as.numeric(study_dates[args[3]]), format = "%Y-%m-%d", origin = "1970-01-01")
+
+df_input <- read_feather(
+  here::here("output", paste0("input_processed_", args[[1]], "_", 
+    year(study_start_date), "_", year(study_end_date), ".arrow")))
+
+lab <- ifelse(args[1] == "infants", "Age (Months)", 
+       ifelse(args[1] == "infants_subgroup", "Age (Months)", "Age (Years)"))
+
+plot_age <- ggplot(data = df_input, aes(age, frequency(age))) + geom_col(width = 0.9) +
+  xlab(lab) + ylab("Frequency")
+
+ggsave(
+  plot = plot_age,
+  filename = paste0("descriptive_", args[[1]], "_", year(study_start_date),
+    "_", year(study_end_date), ".png"), path = here::here("output"),
+)

project.yaml

@@ -22,7 +22,14 @@ actions:
     outputs:
       highly_sensitive:
         cohort: output/input_processed_older_adults_2016_2017.arrow
-
+
+  describe_dataset_older_adults_s1:
+    run: r:latest analysis/report.R older_adults season1_start_date season1_end_date
+    needs: [process_dataset_older_adults_s1]
+    outputs:
+      moderately_sensitive:
+        cohort: output/descriptive_older_adults_2016_2017.png
+
   generate_dataset_adults_s1:
     run: >
       ehrql:v1 generate-dataset analysis/dataset_definition.py