v1.2.0 - 2025-03-26

Add tests, change from default inputs, rename some variables, and add a method which allows the solution of ReplicationsAlgorithm to be less than the initial_replications set.

Added

• Add unit tests for ReplicationsAlgorithm when only 2 replications are run, and for the new find_position() method.
• Add back test for scenario analysis.

Changed

• Linting GitHub action no longer triggers on pull requests.
• Renamed count_unseen and q_time_unseen to be resource-specific (i.e.g count_unseen_nurse).
• Set default alpha to 0.05 for OnlineStatistics.
• Accept instance of Param class as input for run_scenarios() rather than a dictionary.

Fixed

• Add find_position() method to ReplicationsAlgorithm, allowing us to correct results if the solution was below the initial_replications set.

v1.1.0 - 2025-03-21

Add a bash script for executing notebooks and new tests for warm-up patients and replication consistency. Fixes were made to correct notebook configurations, adjust test parameters, refine the replication algorithm, and ensure accurate calculations, such as correcting nurse time usage and setting appropriate values in the replication algorithm. Other changes include updates to default parameters and documentation.

Added

• Bash script to execute all notebooks (with nbconvert add to environment for this).
• Add test related to inclusion/exclusion of warm-up patients in certain metrics (test_warmup_high_demand()).
• Add test for consistent nreps between replications methods.

Changed

• Changed default warm-up length in Param().
• Add pylint line limit so it adheres to PEP-8.
• Allow input of Param() to the objects in replications.py so we can specify parameter set for back tests (so they don't just change results when we change the model defaults).
• Add specific parameters to generate_exp_results.ipynb for back tests.
• Add acknowledgements to README and docstrings.
• Lowered the default min_rep for both confidence_interval_method functions.

Fixed

• Correct choosing_warmup.ipynb to use multiple replications and run length at least 5-10x actual.
• Fix test_waiting_time_utilisation()
• Correct test_klimit() to actually use the parametrize inputs.
• Add correction to nurse_time_used for when patients span the warm-up and data collection period.
• Allow None for the dashed line in plotly_confidence_interval_method().
• Allow a solution below initial_replications in ReplicationsAlgorithm.
• Set target_met back to 0 in ReplicationsAlgorithm if precision is no longer achieved.
• Set test_consistent_outputs() to have no lookahead.

v1.0.0 - 2025-02-27

Lots and lots of changes! Many of these are a result of comments from peer review of code by Tom Monks.

Added

• Virtual environment alternative to conda.
• Bash script to lint repository.
• Lots of new unit tests and functional tests!
• GitHub actions to run tests and lint repository.
• Add MonitoredResource and alternative warm-up results collection.
• Time-weighted statistics - including relevant code (replications.py), documentation (choosing_replications.ipynb), and tests (_replications in tests).
• User-controlled interactive histogram of results in analysis.ipynb.
• Add metrics for unseen patients.

Changed

• Changes to code and environment to accomodate new features (described in 'Added').
• Import simulation as a local package.
• Save all tables and figures.
• Add Tom Monks to author list.
• Expanded README.
• Renaming classes and variables (e.g. Trial to Runner, Defaults to Param).
• Improved log formatting.
• Moved methods (e.g. from analysis.ipynb to simulation/).
• Re-arranged tests into unit tests, back tests and functional tests.

Fixed

• First arrival no longer at time 0.
• Begin interval audit at start of data collection period (rather than start of warm-up period).
• Correct logging message where wrong time was used.
• Add error handling for invalid cores (in model + test) and error message for attempts to log when in parallel.
• Resolved runtime warning with handling for variance 0 in summary_stats().
• Prevent output of standard deviation or confidence intervals from summary_stats() when n<3.
• Add error handling for results processing when there are no arrivals.
• Add error handling for invalid mean in Exponential.

v0.1.0 - 2025-01-09

🌱 First release of the python DES template.