Data

Tools

Indexes

Applications

Experiments

Open Source Science

Recent Releases of rpops

rpops - PoPS 2.0.1

Added

  • calibrate can now use the Mathews Correlation Coefficient as the summary statistic of to keep or reject parameter sets (@ChrisJones687, #145).

  • validate, calibrate, pops_multirun, and pops now take usesurvivalrates, survivalratemonth, survivalrateday, and survivalratesfile (@ChrisJones687, #147).

    • validate, calibrate, pops_multirun, and pops now take network_movement as a parameter. This parameter controls how dispersal occurs along the network (@ChrisJones687, #147).

Changed

  • validate, calibrate, pops_multirun, auto_manage and pops no longer take networkmindistance and networkmaxdistance as these parameters are now passed in through the parametermeans and parametercov_matrix parameters and are calibrated as part of the calibration if network kernel is selected (@ChrisJones687, #140).

  • calibrate now calibrates the networkmindistance and networkmaxdistance parameters during calibration and they are now part of the parametermeans and parametercov_matrix that are exported from the calibration (@ChrisJones687, #140).

  • calibrate has more flexible success metric options removes usedistance, usermse, and usemcc parameters and replaces it with the more flexible successmetrics parameter. Users can now select multiple combinations of different success metrics for the calibration (@ChrisJones687, #150).

  • pops_multirun removed the ability to write all simulations. (@ChrisJones687, #144).

  • the deterministic parameter has been renamed to dispersalstochasticity. This was done to be more consistent with generatestochasticity, movementstochasticity, and establishmentstochasticity parameters (@ChrisJones687, #147).

  • validate, calibrate, pops_multirun, and pops now propogate uncertainty from host and initial conditions. This adds the parameters useinitialconditionuncertainty and usehostuncertainty. If useinitialconditionuncertainty is TRUE the infectedfile and/or exposedfile need to have 2 layers a mean and standard deviation. If usehostuncertainty is is TRUE the host_file needs to have 2 layers a mean and standard deviation. For each model run a host and/or initial conditions are drawn from the mean and sd layers so that each run has a unique host and/or initial conditions this will allow for both propogation of uncertainty from these sources but also partitioning (@ChrisJones687, #151).

Fixed

  • fixed error on parameter draws in select situations (@ChrisJones687, #146).

- R
Published by ChrisJones687 over 3 years ago

rpops - PoPS 2.0.0

Added

  • quantity_allocation_disagreement now takes in the use_distance parameter which is FALSE by default. This allows the model to compute the minimum total distance between observed and simulated infestations (@ChrisJones687, #130).

  • validate, calibrate, pops_multirun, auto_manage and pops now take in networkmindistance, networkmaxdistance, and networkfilename parameters these are used when the anthropogenickernel_type = "network". This allows directed spread along a network such as a railroad (@ChrisJones687 and @wenzeslaus, #131)

Changed

  • validate now takes the variable pointfile and uses it to calculate statistics based on the pointfile in addition to the raster file also calculates new measures of model performance accuracy, precision, recall, and specificity (@ChrisJones687, #124).

  • calibrate no longer uses success_metrics and checks parameters as these are both handled internally by auto-updating if the values are too far of in the first generation. This makes for a simpler and faster user experience (@ChrisJones687, #130).

  • validate no longer uses successmetrics parameter but now has the added parameters usermse and usedistance that are used in `quantityallocation_disagreement`. This allows for a more intuitive user interface (@ChrisJones687, #130).

  • quantity_allocation_disagreement variable configuration changed to use_configuration to be more consistent with variable names. (@ChrisJones687, #130).

  • pops_multirun can now use variable writeoutput = "allsimulations" to write out susceptible, exposed, and infected rasters from all simulations. This will allow for a future update where simulations can easily be started from previous simulation outputs. (@ChrisJones687, #133)

- R
Published by ChrisJones687 about 4 years ago

rpops - PoPS 1.1.0

Added

  • New dispersal kernels added: Uniform, Power-law, Deterministic neighbor, Hyperbolic-Secant, Gamma, Weibull, Normal, and Logistic (@ChrisJones687, #73).

    • These kernels are usable for both (natural_dispersal_kernel, anthropogenic_dispersal_kernel).
    • Used in the following functions (pops, pops_multirun, auto_manage, calibrate, and validate).

  • Overpopulation module added to pops-core (@wenzeslaus, #83).

  • Spatial Index to increase computational speed (@ChrisJones, #67)

Changed

  • Exposed and Infected populations can both be present at the start of a simulation (@ChrisJones687, #92).

  • Internal functions for data handling have switched from the raster package to the terra package (@ChrisJones687, #79).

    • Adds terra dependency.
    • Greatly speeds up data handling and preparation.

  • Mask parameter now used in pops-multirun for post processing data for visualization (@ChrisJones687, #104).

  • Calibration function now takes verbose parameter (@nfkruska, #99).

  • All functions now support vrt data types (@ChrisJones687, #97).

  • Raster files can now be read from S3 buckets (@chrisJones687, #75).

    • needed for model-api for dashboard

  • Outputs can now be saved with write_outputs and output_folder_path parameters (@ChrisJones687, #111).

  • Host map now used as initial mask for post processing and validation calculations (@ChrisJones687, #112).

  • Movements now moves exposed, resistant, and mortality tracked populations (@ChrisJones687, #118).

  • Mortality can now occur at various timesteps not just yearly (@ChrisJones687, #118).

    • Can be either "day", "week", "month", "year", or "everynsteps".
    • adds parameters mortality_frequency and mortality_frequency_n.

  • Validation now exports the statistics for each output and the cumulative statistics for each year (@ChrisJones687, #121).

  • Treatments now update total_hosts (@ChrisJones687, #122)

- R
Published by ChrisJones687 over 4 years ago

rpops - PoPS 1.0.2

Version 1.0.2 of the PoPS Core C++ library and its interfaces: rpops R package and r.pops.spread GRASS GIS module. The release of rpops includes:

  • Susceptible-infected (SI) and susceptible-exposed-infected (SEI) host phases (model_type, latency_period),
  • Host mortality tracking (mortality_rate, mortality),
  • Host removal and pesticide application treatments (treatments, treatment_date, pesticide_duration),
  • Host resistance based on pesticide application treatments (pesticide_duration > 0),
  • Treatments applied only to a ratio of hosts (treatment_application),
  • Yearly pest removal based on lethal temperature (lethal_temperature, lethal_month),
  • Two different dispersal kernels (natural_dispersal_kernel, anthropogenic_dispersal_kernel),
  • Cauchy and exponential radial dispersal kernels use Von Mises distribution,
  • Seasonal spread (seasonality in months),
  • Multiple stochastic runs (pops_multirun),
  • Parallel execution of multiple runs (number_of_cores in pops_multirun),
  • Output of average and standard deviation of infected hosts across multiple runs and for a single stochastic run (simulation_mean, single_run, simulation_sd),
  • Average and standard deviations for output averages (number_infecteds, infected_areas),
  • Infection probability output in percent (probability),
  • Spread rate measurement in 4 cardinal directions (west_rate, east_rate, south_rate, north_rate ),
  • Distance to quarantine in 4 cardinal directions (north_distance_to_quarantine, south_distance_to_quarantine, east_distance_to_quarantine, west_distance_to_quarantine ),
  • Probability of quarantine escape (escape_probability).

Version 1.0.2 Changes * Multirun and validate draw a parameter set from the distribution for each run as intended * Exposed populations now exported each time-step * Fixed error when calibrating for less than 4 parameters

- R
Published by ChrisJones687 over 5 years ago

rpops - PoPS 1.0.0

Version 1.0.0 of the PoPS Core C++ library and its interfaces: rpops R package and r.pops.spread GRASS GIS module. The release of rpops includes:

  • Susceptible-infected (SI) and susceptible-exposed-infected (SEI) host phases (model_type, latency_period),
  • Host mortality tracking (mortality_rate, mortality),
  • Host removal and pesticide application treatments (treatments, treatment_date, pesticide_duration),
  • Host resistance based on pesticide application treatments (pesticide_duration > 0),
  • Treatments applied only to a ratio of hosts (treatment_application),
  • Yearly pest removal based on lethal temperature (lethal_temperature, lethal_month),
  • Two different dispersal kernels (natural_dispersal_kernel, anthropogenic_dispersal_kernel),
  • Cauchy and exponential radial dispersal kernels use Von Mises distribution,
  • Seasonal spread (seasonality in months),
  • Multiple stochastic runs (pops_multirun),
  • Parallel execution of multiple runs (number_of_cores in pops_multirun),
  • Output of average and standard deviation of infected hosts across multiple runs and for a single stochastic run (simulation_mean, single_run, simulation_sd),
  • Average and standard deviations for output averages (number_infecteds, infected_areas),
  • Infection probability output in percent (probability),
  • Spread rate measurement in 4 cardinal directions (west_rate, east_rate, south_rate, north_rate ),
  • Distance to quarantine in 4 cardinal directions (north_distance_to_quarantine, south_distance_to_quarantine, east_distance_to_quarantine, west_distance_to_quarantine ),
  • Probability of quarantine escape (escape_probability).

- R
Published by wenzeslaus over 5 years ago

rpops - Beta 0.1

- R
Published by ChrisJones687 about 6 years ago