Issue 12: Add package website (#19)

* Add basic roxygen2 families to functions

* Switch to using enwtheme (I think this is a shortcut for what was written here anyway, but good to use it)

* Add function group descriptions

* Add docs to gitignore

* Add pkgdown to Rbuildignore

* Drop Sam Abbott related content from yml

* Add newline to end of yml

Former-commit-id: ff88f93
Former-commit-id: d93904d1ead1ad620a9fd6045469852fc98efacd

Author: athowes

.Rbuildignore

@@ -10,3 +10,4 @@
 ^docs$
 ^data-raw$
 ^CITATION\.cff$
+^pkgdown$

.gitignore

@@ -21,4 +21,5 @@ test_*.png
 data/models/*doublecensor
 data/models/*full
 data/models/*integrate
-data/models/*reparam
+data/models/*reparam
+docs

DESCRIPTION

@@ -21,7 +21,7 @@ Description: Understanding and accurately estimating epidemiological delay
 License: MIT + file LICENSE
 Encoding: UTF-8
 Roxygen: list(markdown = TRUE)
-RoxygenNote: 7.2.3
+RoxygenNote: 7.3.1
 Imports: 
     brms,
     cmdstanr,

R/fitting-and-postprocessing.R

@@ -1,4 +1,6 @@
 #' Sample from the posterior of a model with additional diagnositics
+#' 
+#' @family postprocess
 #' @export
 sample_model <- function(model, data, scenario = data.table::data.table(id = 1),
                          diagnostics = TRUE, ...) {
@@ -48,6 +50,8 @@ sample_model <- function(model, data, scenario = data.table::data.table(id = 1),
 }
 
 #' Sample from the posterior of an epinowcast model with additional diagnositics
+#' 
+#' @family postprocess
 #' @export
 sample_epinowcast_model <- function(
   model, data, scenario = data.table::data.table(id = 1),
@@ -115,6 +119,8 @@ sample_epinowcast_model <- function(
 }
 
 #' Add natural scale summary parameters for a lognormal distribution
+#' 
+#' @family postprocess
 #' @export
 add_natural_scale_mean_sd <- function(dt) {
   nat_dt <- data.table::copy(dt)
@@ -127,6 +133,8 @@ add_natural_scale_mean_sd <- function(dt) {
 }
 
 #' Extract posterior samples for a lognormal brms model
+#' 
+#' @family postprocess
 #' @export
 #' @importFrom posterior as_draws_df
 extract_lognormal_draws <- function(
@@ -164,6 +172,8 @@ extract_lognormal_draws <- function(
 }
 
 #' Extract posterior samples for a lognormal epinowcast model
+#' 
+#' @family postprocess
 #' @export
 extract_epinowcast_draws <- function(
   data, id_vars, from_dt = FALSE
@@ -199,6 +209,8 @@ extract_epinowcast_draws <- function(
 }
 
 #' Primary event bias correction
+#' 
+#' @family postprocess
 #' @export
 primary_censoring_bias_correction <- function(draws) {
   draws <- data.table::copy(draws)
@@ -210,6 +222,8 @@ primary_censoring_bias_correction <- function(draws) {
 }
 
 #' Convert posterior lognormal samples to long format
+#' 
+#' @family postprocess
 #' @export
 draws_to_long <- function(draws) {
   long_draws <- data.table::melt(
@@ -221,6 +235,8 @@ draws_to_long <- function(draws) {
 }
 
 #' Make posterior lognormal samples relative to true values
+#' 
+#' @family postprocess
 #' @export
 make_relative_to_truth <- function(draws, secondary_dist, by = "parameter") {
   draws <- merge(
@@ -238,6 +254,8 @@ make_relative_to_truth <- function(draws, secondary_dist, by = "parameter") {
 #' @param not_by A vector of columns to exclude from the grouping
 #' This will be overridden if by is specified.
 #' @inheritParams summarise_variable
+#' 
+#' @family postprocess
 #' @export
 summarise_draws <- function(draws, sf, not_by = "value", by) {
   if (missing(by)) {
@@ -280,6 +298,8 @@ summarise_draws <- function(draws, sf, not_by = "value", by) {
 #' @param sf The number of significant figures to use
 #' @param variable The variable to summarise
 #' @param by A vector of columns to group by
+#' 
+#' @family postprocess
 #' @export
 summarise_variable <- function(draws, variable, sf = 6, by = c()) {
   if (missing(variable)) {

R/models.R

@@ -1,4 +1,6 @@
 #' Estimate delays naively
+#' 
+#' @family model
 #' @export
 naive_delay <- function(formula = brms::bf(delay_daily ~ 1, sigma ~ 1), data,
                         fn = brms::brm, family = "lognormal", ...) {
@@ -10,6 +12,8 @@ naive_delay <- function(formula = brms::bf(delay_daily ~ 1, sigma ~ 1), data,
 }
 
 #' Estimate delays with filtering of the most recent data
+#' 
+#' @family model
 #' @export
 filtered_naive_delay <- function(
   formula = brms::bf(delay_daily ~ 1, sigma ~ 1), data, fn = brms::brm,
@@ -26,6 +30,8 @@ filtered_naive_delay <- function(
 }
 
 #' Estimate delays adjusted for censoring
+#' 
+#' @family model
 #' @export
 censoring_adjusted_delay <- function(
   formula = brms::bf(
@@ -39,6 +45,8 @@ censoring_adjusted_delay <- function(
 
 #' Estimate delays adjusted forcensoring using a
 #' latent model
+#' 
+#' @family model
 #' @export
 latent_censoring_adjusted_delay <- function(
   formula = brms::bf(
@@ -103,6 +111,8 @@ latent_censoring_adjusted_delay <- function(
 
 #' Estimate delays with filtering of the most recent data and
 #' censoring adjustment
+#' 
+#' @family model
 #' @export
 filtered_censoring_adjusted_delay <- function(
   formula = brms::bf(
@@ -120,6 +130,8 @@ filtered_censoring_adjusted_delay <- function(
 }
 
 #' Estimate delays adjusted for right truncation
+#' 
+#' @family model
 #' @export
 truncation_adjusted_delay <- function(
   formula = brms::bf(
@@ -134,6 +146,8 @@ truncation_adjusted_delay <- function(
 }
 
 #' Estimate delays adjusted for censoring and right truncation
+#' 
+#' @family model
 #' @export
 truncation_censoring_adjusted_delay <- function(
   formula = brms::bf(
@@ -151,6 +165,8 @@ truncation_censoring_adjusted_delay <- function(
 
 #' Estimate delays adjusted for right truncation and censoring using a
 #' latent model
+#' 
+#' @family model
 #' @export
 latent_truncation_censoring_adjusted_delay <- function(
   formula = brms::bf(
@@ -256,6 +272,8 @@ latent_truncation_censoring_adjusted_delay <- function(
 #' Estimate delays from the backward delay distribution + brms
 #' @param data_cases data frame consisting of integer time column and incidence
 #' column
+#' 
+#' @family model
 #' @export
 dynamical_censoring_adjusted_delay <- function(
     formula = brms::bf(
@@ -407,11 +425,13 @@ dynamical_censoring_adjusted_delay <- function(
   return(fit)
 }
 
+#' @family model
 #' @export
 dynamical_censoring_adjusted_delay_wrapper <- function(data, data_cases, ...) {
   dynamical_censoring_adjusted_delay(data = data, ...)
 }
 
+#' @family model
 #' @export
 epinowcast_delay <- function(formula = ~ 1, data, by = c(),
                              family = "lognormal", max_delay = 30,

R/observe.R

@@ -1,4 +1,6 @@
 #' Observation process for primary and secondary events
+#' 
+#' @family observe
 #' @export
 observe_process <- function(linelist) {
   clinelist <- data.table::copy(linelist)
@@ -23,6 +25,8 @@ observe_process <- function(linelist) {
 }
 
 #' Filter observations based on a observation time of secondary events
+#' 
+#' @family observe
 #' @export
 filter_obs_by_obs_time <- function(linelist, obs_time) {
   truncated_linelist <- data.table::copy(linelist)
@@ -36,6 +40,8 @@ filter_obs_by_obs_time <- function(linelist, obs_time) {
 }
 
 #' Filter observations based on the observation time of primary events
+#' 
+#' @family observe
 #' @export
 filter_obs_by_ptime <- function(linelist, obs_time,
                                 obs_at = c("obs_secondary", "max_secondary")) {
@@ -67,6 +73,8 @@ filter_obs_by_ptime <- function(linelist, obs_time,
 }
 
 #' Pad zero observations as unstable in a lognormal distribution
+#' 
+#' @family observe
 #' @export
 pad_zero <- function(data, pad = 1e-3) {
   data <- data.table::copy(data)
@@ -79,6 +87,8 @@ pad_zero <- function(data, pad = 1e-3) {
 }
 
 #' Drop zero observations as unstable in a lognormal distribution
+#' 
+#' @family observe
 #' @export
 drop_zero <- function(data) {
   data <- data.table::copy(data)

R/plot-helpers.R

@@ -1,4 +1,6 @@
 #' Calculate the cohort-based or cumulative mean
+#' 
+#' @family plot
 #' @export
 calculate_cohort_mean <- function(data, type = c("cohort", "cumulative"),
                                   by = c(), obs_at) {
@@ -24,6 +26,8 @@ calculate_cohort_mean <- function(data, type = c("cohort", "cumulative"),
 }
 
 #' Calculate the truncated mean by observation horizon
+#' 
+#' @family plot
 #' @export
 calculate_truncated_means <- function(draws, obs_at, ptime,
                                       distribution = function(x, y, z) {

R/plot.R

@@ -1,4 +1,6 @@
 #' Plot the posterior estimates as densities
+#' 
+#' @family plot
 #' @export
 plot_recovery <- function(data, alpha = 0.8,
                                    quantiles = c(0.05, 0.35, 0.65, 0.95), ...) {
@@ -13,6 +15,8 @@ plot_recovery <- function(data, alpha = 0.8,
 }
 
 #' Plot the relative difference between true values and posterior estimates
+#' 
+#' @family plot
 #' @export
 plot_relative_recovery <- function(relative_data, alpha = 0.8,
                                    quantiles = c(0.05, 0.35, 0.65, 0.95), ...) {
@@ -26,6 +30,8 @@ plot_relative_recovery <- function(relative_data, alpha = 0.8,
 }
 
 #' Plot cases by observation window
+#' 
+#' @family plot
 #' @export
 plot_cases_by_obs_window <- function(cases) {
   cases[case_type == "primary"] |>
@@ -50,6 +56,8 @@ plot_cases_by_obs_window <- function(cases) {
 }
 
 #' Plot the empirical delay distribution
+#' 
+#' @family plot
 #' @export
 plot_empirical_delay <- function(cases, meanlog, sdlog) {
   plot <- cases |>
@@ -80,6 +88,8 @@ plot_empirical_delay <- function(cases, meanlog, sdlog) {
 }
 
 #' Plot the mean difference between continuous and discrete event time
+#' 
+#' @family plot
 #' @export
 plot_censor_delay <- function(censor_delay) {
   ggplot(censor_delay) +
@@ -94,6 +104,7 @@ plot_censor_delay <- function(censor_delay) {
 }
 
 #' plot empirical cohort-based or cumulative mean vs posterior mean
+#' 
 #' @param summarised_mean Summarised mean as produced by [summarise_variable()]
 #' @param data data used for object fitting
 #' @param truncate account for truncation?
@@ -103,6 +114,8 @@ plot_censor_delay <- function(censor_delay) {
 #' @param ribbon_bounds Bounds of the quantile ribbon. Defaults to 
 #' `c(0.05, 0.95)` which corresponds to the 90% credible interval.
 #' @param ... Additional arguments passed to [ggplot2::aes()].
+#' 
+#' @family plot
 #' @export
 plot_mean_posterior_pred <- function(summarised_mean, obs_mean,
                                      alpha = 0.3, mean = FALSE, ribbon = TRUE,
@@ -149,6 +162,8 @@ make_ribbon_bound <- function(quantile) {
 }
 
 #' Plot empirical cohort-based or cumulative mean
+#' 
+#' @family plot
 #' @export
 plot_cohort_mean <- function(data) {
   gplot <- ggplot(data) +

R/preprocess.R

@@ -1,4 +1,6 @@
 #' For a target variable convert from individual data to counts
+#' 
+#' @family preprocess
 #' @export
 linelist_to_counts <- function(linelist, target_time = "ptime_daily",
                                additional_by = c(), pad_zeros = FALSE) {
@@ -22,6 +24,8 @@ linelist_to_counts <- function(linelist, target_time = "ptime_daily",
 }
 
 #' Convert primary and secondary observations to counts in long format
+#' 
+#' @family preprocess
 #' @export
 linelist_to_cases <- function(linelist) {
   primary_cases <- linelist_to_counts(linelist)
@@ -42,6 +46,8 @@ linelist_to_cases <- function(linelist) {
 }
 
 #' For the observation observed at variable reverse the factor ordering
+#' 
+#' @family preprocess
 #' @export
 reverse_obs_at <- function(dt) {
   dt_rev <- data.table::copy(dt)
@@ -53,6 +59,8 @@ reverse_obs_at <- function(dt) {
 }
 
 #' Construct case counts by observation window based on secondary observations
+#' 
+#' @family preprocess
 #' @export
 construct_cases_by_obs_window <- function(linelist, windows = c(25, 45),
   obs_type = c("stime", "ptime"), upper_window = max(linelist$stime_daily)) {
@@ -94,6 +102,8 @@ construct_cases_by_obs_window <- function(linelist, windows = c(25, 45),
 }
 
 #' Combine truncated and fully observed observations
+#' 
+#' @family preprocess
 #' @export
 combine_obs <- function(truncated_obs, obs) {
   cobs <- rbind(
@@ -105,6 +115,8 @@ combine_obs <- function(truncated_obs, obs) {
 }
 
 #' Calculate the mean difference between continuous and discrete event time
+#' 
+#' @family preprocess
 #' @export
 calculate_censor_delay <- function(truncated_obs, additional_by = c()) {
   truncated_obs_psumm <- data.table::copy(truncated_obs)
@@ -137,6 +149,8 @@ calculate_censor_delay <- function(truncated_obs, additional_by = c()) {
 }
 
 #' Convert from event based to incidence based data
+#' 
+#' @family preprocess
 #' @export
 event_to_incidence <- function(data, by = c()) {
   dd <- data.table::copy(data)