NEWS

swash 3.0.3 (2026-11-01)

breaks_growth() and plot(breaksgrowth) now treat results with less than two segments (which are rare) correctly
calculate_rollmean(infpan) and calculate_rollsum(infpan): Check of the specified rolling window k with respect to the number of time points in the infpan object
Check for NA values in observed and expected vectors in binary_metrics()
Check for regions without any infection cases when using the Swash-Backwash Model with swash_backwash() or swash(infpan)
Corrections in RD documentations

Comprehensive and consistent validation of user-supplied time-series and panel data across all functions and methods that utilize time series; included impute_missing_values option
exponential growth(): Doubling time is set to NA if the estimated growth rate r is non-positive
If no start values for NLS estimation is specified in exponential_growth(), coefficients from the OLS estimation are used as start values (as in logistic_growth())

Fixed bugs in balancing datasets procedure in load_infections_paneldata(), is_balanced(), and as_balanced()
Checking of correct time format of 'col_date' in load_infections_paneldata() and swash_backwash()
Fixed bug in bootstrap sampling in confint(sbm)
A potential error in plot_coef_ci() involving exclusively positive or exclusively negative significant estimates has been fixed
Fixed bug in dealing with NA values in logistic_growth(), exponential_growth(), and breaks_growth()
Fixed bug in bisection mode in logistic_growth()
Checking of specified parameters in confint(sbm)
Checking 'infections' vector length and type of parameter 'GP' in R_t()
Restored erroneously removed NEWS.md

Corrections in NAMESPACE (required for correct package installation)
exponential growth(): R0 is set to NA (instead of 0) if the estimated growth rate is negative
metrics(): Correcting R-squared for zero variance, added na.rm parameter (default: TRUE)

Implemented testthat testing in tests/ folder, keeping former swash_test.R in examples/ folder
Extensions of documentation

Complete change to the nbmatrix() function: creation of an instance of the new nbmatrix class
Replacement of the old nbstat() function with the nbstat() method of the nbmatrix class
Rearrangement of the code in modules for better readability and easier maintaining

Spatial statistics based on neighborhood matrix: Global Getis-Ord, Global Moran's I,
Local Getis-Ord Gi*, with all of them being methods of class nbmatrix
Plotting a map from an nbmatrix object
Importing geodata (sf) in infpan objects
Plotting maps of attributes in an infpan object with method plot_map()

Analyses are conducted via infpan objects rather than sbm objects
Former method plot_regions() for class sbm is replaced by generic method plot() for infpan objects
Former methods growth() and growth_initial() for sbm objects are are now methods for the infpan class
Former function plot_breakpoints() is replaced by function breaks_growth() and the corresponding plot() method

Importing infections panel data via load_infections_paneldata(), which creates and instance of the new class infpan
Calculation of spread indicators for infpan objects such as incidence and effective reproduction number $R_t$
Function hawkes_growth() for parametrization of a Hawkes process equation for infections
Method growth_hawkes() for parametrization of Hawkes process equations based on infpan objects
Additional NLS estimation in function exponential_growth()
Option to add a constant (if values of y equal to zero occur) in functions logistic_growth() and exponential_growth(), as well as in methods growth(infpan) and growth_initial(infpan)

Functions metrics(), binary_metrics() and binary_metrics_glm() now return results invisible
Check for equal length of input vectors in logistic_growth()