High-Performance Open-Source Archive
config(), breaking the components of a
config object out into a hierarchy of (new) S3 objects:
predictors, life_cycle,
parameters, predictor_spec,
transition_function, and transition. This has
numerous advantages, including the ability for users to build up
individual components of a config with validation from the
new classes, rather than constructing the entire config
object in one step. This update means users need to modify any existing
config objects to fit the new schema. The most significant
changes are:
life_cycle class,
which contains a list of transition objects. Previously,
this data was represented in the transitions and
parameters attributes of a config.transition_function. Previously,
configs contained strings with the name of functions to
get() from the environment; now we pass the
transition_functions themselves. This removes the model’s
dependence on the global R environment, improving reproducibility.predictor_spec and parameters,
passing parameter and predictor values to
transition_functions is now done more explicitly.configs to conform to the new
schema.read_config() and write_config(),
which provided the ability to serialize and deserialize
configs. These functions were not sufficient to
comprehensively serialize all configs for reproducible
model results. Now, the package only handles in-memory objects.run_all_configs() and the related
parallel package dependency. This function was just a thin
wrapper over a call to lapply or a parallel apply
operation.graph_lifecycle() for visualizing the flow
between life stages, and the dependency on the package
igraph. Instead, life_cycle includes a print
method with useful information on the flow between life stages.vary_param() and
vary_many_params().graph_population_each_group() and
graph_population_overall_trend(). These functions were
intended as an easy way to visualize model results, but they did not
handle all cases, since model output can be highly variable. We now
recommend users write their own code, e.g. with ggplot2, to
visualize results.run(): age_group, process and
infected. These columns were populated based on the
assumption that life stage names were structured as a three-character
string like
<process><infected><age_group>, and used
downstream in the graphing functions. Now, users are completely free to
name life stages as they wish.annual_growth_rate(), for determining the annual
factor by which population changes. Results from the
growth_rate() function can be sensitive to the specific
time period being modeled – this aims to be a more universally
applicable alternative.growth_rate() to use base R.configs to the package.
ogden2005, which replicates an existing deer tick
population model by Ogden et al. (2005).winter_tick, based on literature on the winter tick and
used as an example of the package’s flexibility for studying different
tick species.host_example_config,
infect_example_config, and
temp_example_config, which are used as examples of package
use. Added examples with these configs to the readme.config() to take a single
predictors argument, instead of separate
weather and host_comm arguments. This
simplifies the config() structure, and allows using other
types of data as a predictor for transitions.
write_config() arguments are updated to reflect this
change.snow_cover_fun() transition function which is
used in the winter_tick model configuration.write_config() to write steps and initial
population as integers rather than decimal numbers.IxPopDyMod package
Need mirroring services?
Contact our team at info@vpspulse.com.
Mirror powered by VPSpulse
Infrastructure sponsored by VPSPulse & Secure Payments by ArionPay.