Skip to content

Latest commit

 

History

History
36 lines (29 loc) · 4.72 KB

SMG.md

File metadata and controls

36 lines (29 loc) · 4.72 KB
Raw

System Maintenance Guide

Quickstart

Adding a new location

  1. The data for the location needs to be available via epiforecasts/covidregionaldata - check the system maintenance guide for more information

  2. Add your location to the list in R/lists/dataset-list.R (in alphabetical order), setting stable=FALSE until testing is complete. Ensure the key is the name. Ensure the name is unique and isn't repeated in the collated-derivative-list (this breaks publishing).

    Region$new(name = "middle-earth", stable=FALSE),

  3. Run it! Rscript R/run-region-updates.R -w -u -i middle-earth/* (executing in very verbose mode, including unstable locations, only include the new location and all sub-locations)

    This should take in the order of (n*80)/cores minutes where n is the number of sub-locations to process

Adding a new collated derivative

  1. Add the collation to R/lists/collated-derivative-list.R - this should be in a similar style to Regions and an existing one should be fairly self explanatory. Ensure the name is unique and isn't repeated in the dataset-list (this breaks publishing).

Region - additional control

The region object allows for a range of values to be specified. There are slight differences between superregion and region - the "Object" column below highlights any differences

It is possible to nest multiple different datasets for the same location - check the folder_name and dataset_folder_name options below for examples.

Property Object Mandatory Default Purpose Example
name All Yes - this will be treated as the name used in any file path and is the default for covid_regional_data_identifier name = "germany"
covid_regional_data_identifier

covid_national_data_identifier		 Region

SuperRegion		 No name

"ecdc"		 Used for the call to covidregionaldata::get_regional_data / covidregionaldata::get_national_data to specify the country / source parameter. Used to shim inconsistencies between the two libraries (e.g. united-kingdom / UK ). Name differs between region / superregion. covid_regional_data_identifier = "UK"
case_modifier All No NA A lambda that modifies the cases object. This is expected to return the cases object. It allows for additional filtering or modifying of the source data if needed. This should be used with caution as it provides a method of "tinkering" with the source data with potential loss of data integrity. case_modifier = function(cases) { ... return(cases)}
generation_time All No loads "data/generation_time.rds" Optionally provide alternative data object to replace that loaded from the generic generation_time.rds file generation_time = readRDS(here::here("data", "alternative_generation_time.rds"))
incubation_period All No loads "data/incubation_period.rds" Optionally provide alternative data object to replace that loaded from the generic incubation_period.rds file incubation_period = readRDS(here::here("data", "alternative_incubation.rds"))
reporting_delay All No loads "data/onset_to_admission_delay.rds" Optionally provide alternative data object to replace that loaded from the generic onset_to_admission_delay.rds file reporting_delay = readRDS(here::here("data", "alternative_delay.rds"))
cases_subregion_source Region No "level_1_region" If the columns returned by covidregionaldata are not using the standard naming this can be reused to map the correct column for region cases_subregion_source = ...
data_args Region No NULL Optional extra arguments to hand to the get_regional_data method data_args = list(nhsregions = TRUE)
region_scale Region

SuperRegion		 No "Region"

"Country"		 Used to refer to the region in report. E.g. "State" for USA region_scale = "State"
stable All No TRUE Controls if it is eligible for inclusion in a full run. Regions under development (or suffering from data issues) can be flagged as stable=FALSE and excluded by default stable = FALSE
folder_name All No NA if specified it replaces the dataset name in the folder structure folder_name="USA"
dataset_folder_name Region No "cases" allows for specifying the dataset is something other than cases. Typically used as a pair with the folder_name flag to co-locate to datasets sensible name="uk-hospital-admissions", folder_name="united-kingdom", dataset_folder_name="hospital-admissions" with another dataset of name="united-kingdom" - this will produce data in subnational/united-kingdom/<cases or hospital-admissions>/...