Convert cumulative counts to incremental counts

Converts cumulative counts (total reported as of each report date) to incremental counts (new cases reported at each report date). This is necessary because some data sources (e.g., epidatr NHSN data) return cumulative counts, but baselinenowcast::as_reporting_triangle() expects incremental counts.

Usage

cumulative_to_incremental(data, group_cols = c("reference_date", "location"))

Arguments

data

A data frame with columns reference_date, report_date, and count. The count column should contain cumulative counts.

group_cols

Character vector of additional columns to group by when computing differences (e.g., "location", "signal"). Default is c("reference_date", "location").

Value

A data frame with the same structure as input, but with count converted to incremental counts (difference from previous report date).

Details

For each reference date (and optionally other grouping columns), the function:

1. Sorts by report date

2. Computes the difference from the previous report date's count

3. Uses 0 as the "previous" count for the first report date

This assumes that the first report date for each reference date represents the initial count (i.e., there was 0 before it).

Examples

# Example with cumulative data
cumulative_data <- data.frame(
  reference_date = as.Date(c("2024-01-06", "2024-01-06", "2024-01-06")),
  report_date = as.Date(c("2024-01-13", "2024-01-20", "2024-01-27")),
  count = c(100, 120, 125),  # cumulative
  location = "ca"
)

incremental_data <- cumulative_to_incremental(cumulative_data)
# count is now: 100, 20, 5 (the differences)