Changelog

Source: NEWS.md

t3 3.0.0 - 2025-04-15

Changed

• Level 1 optimization gather methods 1.1 and 1.2 for raising factors computation
• Fix typo (weight instead of weigth) in :
  • R6 reference object class well,
  • R6 reference object class standardisedsampleset,
  • process 1.3 conversion_weight_categories,
  • process 2.4 well_set_weight_categories(). Warning: you will need to correct this typo in your scripts

Added

• Add method fishing_effort() gathering methods 1.5 to 1.8:
• set_duration : calculated according to a linear function of catch weight with two parameters a and b. These are found through a reference table (set_duration_ref.csv), for each year, ocean, fishing school and country.
• time_at_sea : the process divides the day’s time at sea declared between the activities, allowing the allocation of time at sea, recorded on that date. If no activity to allocate time at sea is recorded on a given date, with a non-zero time at sea, a transit activity is created (whose id_activity contains #666#) to allocate the time ate sea of that date.
• fishing_time : the process module the duration of a working day according to the real sunrise and sunset of each day. It then divides the day’s fishing time between the fishing activities recorded on that date. If no fishing activity is recorded on a given date with a non-zero fishing time, a searching activity is created (whose id_activity contains #666#) to allocate the fishing time of that date.
• searching_time = fishing_time - set_duration.

Removed

• Methods 1.5 to 1.8 and 1.2 (rf2).

Warning: you will need to correct this typo in your scripts

t3 2.1.2 - 2025-04-14

Changed

• Add year 2025 in the set duration referential (set_duration_ref.csv).
• Fix typo (weight instead of weigth) in :
  • R6 reference object class well,
  • R6 reference object class standardisedsampleset,
  • process 1.3 conversion_weight_categories,
  • process 2.4 well_set_weight_categories(). Warning: you will need to correct this typo in your scripts

Added

• Possibility of considering AVDTH and Observe data as input for model fitting at level 3.

t3 2.1.1 - 2025-03-21

Changed

• Manage fishing effort calculation for activities with multiple floating objects or object operation code declared.

t3 2.1.0 - 2025-03-19

Added

• Add a catch_count column in elementarycatches data, recording the number of individuals caught in a set, which can be recorded in place of catch_weight for by-catch species, in the new version of observe database.
  
• Add method (object_model_data$activitycoderefs_data()) to import activity code referential for time allocation (time_allocation_activity_code_ref.csv).
• Add argument species_fate_codes in method object_model_data$activities_object_creation() to import activities and elementarycatch(es) with the same function.
• Add argument activity_code_ref in level_1, process 1.5 set_duration(), 1.6 time_at_sea() and 1.7 fishing_time().

Changed

• Optimization :
  • Change elementary catch(es) type from R6 Object to data.frame included into activity R6 Object.
  • Gather function to create (object_model_data()) and add (full_trips$add_elementarycatches()) activities and elementary catches.
• Update the activity code referential for time allocation (time_allocation_activity_code_ref.csv)
• Improve fishing effort evaluation. The duration of activities is now calculated according to the activity code referential, in order to allocate time only to significant activities:
• set_duration : calculated according to a linear function of catch weight with two parameters a and b. These are found through a reference table (set_duration_ref.csv), for each year, ocean, fishing school and country.
• time_at_sea : the process divides the day’s time at sea declared between the activities, allowing the allocation of time at sea, recorded on that date. If no activity to allocate time at sea is recorded on a given date, with a non-zero time at sea, a transit activity is created (whose id_activity contains #666#) to allocate the time ate sea of that date.
• fishing_time : the process module the duration of a working day according to the real sunrise and sunset of each day. It then divides the day’s fishing time between the fishing activities recorded on that date. If no fishing activity is recorded on a given date with a non-zero fishing time, a searching activity is created (whose id_activity contains #666#) to allocate the fishing time of that date.
• searching_time = fishing_time - set_duration.
• Manage fishing effort calculation for activities with multiple floating objects or object operation code declared.

Removed

• Remove method object_model_data$elementarycatches_object_creation().
• Remove method full_trips$add_elementarycatches().
• Remove elementarycatches and elementarycatch R6 Object definition and methods.

t3 2.0.1 - 2025-01-21

Changed

• Change the type of the flag_codes parameter of object_model_data() function from integer to character (three-letter FAO code(s) for the country(ies)).

Added

• Add arguments country_flag and input_type = "observe_database" by default, in level_3 process 3.4 data_formatting_for_predictions() and 3.5 model_predictions() and in function t3_level3().

t3 2.0.0 - 2024-11-29

Added

• Add data source observe_database.
• Add weight category + 60kg (code 14) for free school, undetermined school and floating object school in Atlantic ocean and Indian ocean.
• Add functionality for querying multiple databases, for example the main and acquisition observe databases to simultaneously import and process ‘recent’ data from acquisition database, not yet imported into the main database, and older data from the main database.
• Add activity code referential to allocate time at sea and fishing time and set duration.

Changed

• Update vignettes.
• Update unit tests.
• Use of codama for checking arguments.
• Update referential tables : set_duration_ref, length_step and length_weight_relationship.

Removed

• Remove data source t3_db.

t3 1.0.0 - 2022-04-21

Added

• Implementation of unit tests
• Implementation of the documentation for the level 3
• Development of output extraction functions for process 1.1
• Adding an output directory initialization function and integrating it into the processes

Changed

• Fix minor bugs and optimization of the code and regarding the overall process
• Update of the level 3 process

t3 0.9.1 - 2020-09-21

Added

• Implementation for the import of annual dataset from outputs of the level 1 and 2 (period length customizable)
• Implementation of bootstrap method as base for all confidence intervals
• Computation of confidence interval (bootstrap interval) for the nominal catches by species and by fishing mode
• Implementation of outputs for the he nominal catches by species and by fishing mode
• Fix minor bugs in all different processes (P1, P2, P3)

Changed

• Improve modelling configuration (addition of new parameters)
• Rewrite several figures for the model checking
• Improve time computing using faster implementation of random forests for the species composition modeling in the level3

Removed

• Disabling task and effort computation while waiting for confident interval implementation
• Remove the documentation from the package. Stay available at https://ob7-ird.github.io/t3/index.html

t3 0.9.0 - 2020-06-05

Added

• Implentation of the model layer in R6class
• Implementation of all logbooks standardization sub-processes
• Implementation of all samples standardization sub-processes
• Implementation of sub-processes for assessing species composition in relation to non-sampled assemblages
• Strict selection of samples used in the model training according to several criteria on sets and wells (school type, location, date, sample quality).
  
• Classification of unknown school type (no data available) by clustering according to major tuna species composition.
  
• Development of a statistical model accounting for:
  • Spatial and temporal structure of the fishery (location and date of the catch)
  • Vessel specificity which impacts the species composition of the catch
  • Crew reporting information which enables more flexibility in the model when a marginal composition is caught
  • Several years of fishing in the training to increase the robustness of the estimates
• Model checking to ensure reliability of the results
• Computation of a confidence interval (bootstrap method) which give the trust level on catch estimations