prep_sim_data: Prepare data for simulation

In aarmiller/delaySim: Package to run delayed diagnosis simulations

Prepare data for simulation

Description

Prepare data for simulation

Usage

prep_sim_data(
  time_map_data,
  by_days = 1,
  start_day = 1,
  time_before = 365,
  event_name = "any_ssd",
  cp_method = "lm_quad",
  specify_cp = NULL,
  set_cp_method = NULL,
  eval_criteria = "AIC",
  week_period = FALSE,
  prediction_bound_for_sim = FALSE,
  auto_reg = FALSE,
  dur_bins = 1:30
)

Arguments

time_map_data	a time_map of visits
by_days	the number of days to aggregate by in counting periods
start_day	When to start counting prior to the index
event_name	The variable name for the event indicator
cp_method	The change-point method to fit visit counts (i.e. "lm","lm_quad","lm_cube", "quad", "cube", "exp", "spline", "cusum", "pettitt", or "set_cp"). "set_cp" is not a change point detection method, rather it is a method to specify the change point and method used to model the data prior to the specified change point
specify_cp	Set a specific change point you want to use instead of searching for optimal change point. Enter a postive integer value repersenting the days before the index on which you you want to specify the change point. (e.g. 100 would be 100 days before the index) This is a required argument if cp_method = "set_cp"
set_cp_method	The method used to model the data prior to a specified change point for the "set_cp" cp_method (i.e. "linear", "cubic", "exponential", etc.)
eval_criteria	The evaluation criteria used to find change points, if using a linear regression method
week_period	Logical to incorporate a "day of the week" effect into the linear model, if method is "pettitt" of "cusum". Note this is only sensible for one-day period aggregation
prediction_bound_for_sim	Logical to specify whether or not to use the estimated predicted value or the upper bound 90 prediction value in the simulations. The defualt is FALSE which uses the estimated predicited value
auto_reg	Logical that determines whether expected counts use a time-series framework that incorporates autoregression. Will automatically fit periodicity, automatically setting week_period to TRUE. Only relevant for cusum and pettitt methods
dur_bins	Numeric vector containing all days for which the simulations should calculate delay duration statistics. If values contained in this vector are greater than the changepoint, they will be omitted from the output.

Examples

## Example to detect change point using cusum ##
out <- prep_sim_data_delay(time_map_data, by_days=1, start_day=1, event_name = "any_ssd", cp_method = "cumsum")

## Examples showing different ways to specify a change point ##

# Using a change point detection method (e.g. cumsum or lm_cube), but enforce a specific change point
cp <- 21
out <- prep_sim_data_delaydx(time_map_data, by_days=1, start_day=1, event_name = "any_ssd", cp_method = "cumsum", specify_cp = cp)

# Using the "set_cp" in the cp_method argument to apply a cubic model to the data prior to the change point
cp <- 21
out <- prep_sim_data_delaydx(time_map_data, by_days=1, start_day=1, event_name = "any_ssd", cp_method = "set_cp", specify_cp = cp,
set_cp_method = "cubic")

aarmiller/delaySim documentation built on Jan. 2, 2023, 11:23 p.m.