eval_metrics {ATQ}

Evaluate Alarm Metrics for Epidemic Models

Description

This function calculates various performance metrics for epidemic alarm systems across different lags and thresholds. It evaluates False Alarm Rate (FAR), Added Days Delayed (ADD), Average Alarm Time Quality (AATQ), First Alarm Time Quality (FATQ), and their weighted versions (WAATQ, WFATQ).

Usage

eval_metrics(data, maxlag = 15, thres = seq(0.1, 0.6, by = 0.05), topt = 14)

Arguments

data

A data frame containing the epidemic data with lagged variables, typically output from the compile_epi function.

maxlag

The maximum number of lags to consider (default: 15).

thres

A vector of threshold values to evaluate (default: seq(0.1, 0.6, by = 0.05)).

topt

Optimal alarm day (default = 14).

Value

A list containing three elements:

metrics

An object of class "alarm_metrics" with the following components:

• FAR: Matrix of False Alarm Rates for each lag and threshold

• ADD: Matrix of Added Days Delayed for each lag and threshold

• AATQ: Matrix of Average Alarm Time Quality for each lag and threshold

• FATQ: Matrix of First Alarm Time Quality for each lag and threshold

• WAATQ: Matrix of Weighted Average Alarm Time Quality for each lag and threshold

• WFATQ: Matrix of Weighted First Alarm Time Quality for each lag and threshold

• best.AATQ: Best model according to AATQ

• best.FATQ: Best model according to FATQ

• best.FAR: Best model according to FAR

• best.ADD: Best model according to ADD

• best.WFATQ: Best model according to WFATQ

• best.WAATQ: Best model according to WAATQ

plot_data

An object of class "alarm_plot_data" for generating plots

summary

An object of class "alarm_metrics_summary" containing summary statistics

See Also

ssir, compile_epi

Examples

# Generate simulated epidemic data
n_rows <- 7421
n_houses <- 1000

epidemic_new <- ssir(n_rows, T = 300, alpha = 0.298, inf_init = 32, rep = 3)

individual_data <- data.frame(
  houseID = rep(1:n_houses, each = ceiling(n_rows / n_houses))[1:n_rows],
  catchID = sample(1:10, n_rows, replace = TRUE),
  schoolID = sample(1:10, n_rows, replace = TRUE),
  num_people = round(rnorm(n_rows, mean = 4, sd = 1)),
  num_elem_child = round(rnorm(n_rows, mean = 1, sd = 1)),
  xStart = 0,
  xEnd = 5,
  yStart = 0,
  yEnd = 5,
  loc.x = rnorm(n_rows, mean = 2.5, sd = 1),
  loc.y = rnorm(n_rows, mean = 2.5, sd = 1),
  individualID = 1:n_rows,
  elem_child_ind = sample(0:1, n_rows, replace = TRUE)
)

compiled_data <- compile_epi(epidemic_new, individual_data)

# Evaluate alarm metrics
alarm_metrics <- eval_metrics(compiled_data,
                              thres = seq(0.1, 0.6, by = 0.05))

# Access the results
summary(alarm_metrics$summary)
alarm_plots <- plot(alarm_metrics$plot_data)

for(i in seq_along(alarm_plots)) { print(alarm_plots[[i]]) }