3. Overview of pelicun features
Coming soon.
This section is under construction.
3.1. Saving/loading samples
All demand, asset, damage, and loss samples can be either computed from other inputs or directly loaded form previously computed and saved samples.
3.2. Logging support
Pelicun produces detailed log files that can be used to document the execution of an assessment as well as information on the host machine and the execution environment. These logs can be useful for debugging purposes. Pelicun emits detailed warnings whenever appropriate, notifying the user of potentially problematic or inconsistent inputs, evaluation settings, or deprecated syntax.
3.3. Uncertainty quantification
Damage and loss estimation is inherently uncertain and treated as a stochastic problem. Uncertainty quantification lies at the core of all computations in pelicun. Pelicun supports a variety of common parametric univariate random variable distributions. With the help of random variable registries, it also supports multivariate distributions, joined with Gaussian copula.
3.4. Assessment types
Pelicun supports scenario-based assessments. That is, losses conditioned on a specific value of an Intensity Measure (IM).
Note
Support for time-based assessments is currently in progress.
3.5. Demand simulation
3.5.1. Model calibration
3.5.2. Sampling methods
3.5.3. RID|PID inference
3.5.4. Sample expansion
3.5.5. Demand cloning
3.6. Damage estimation
3.6.1. Damage processes
3.7. Loss estimation
3.7.1. Loss maps
3.7.2. Active decision variables
3.7.3. Consequence scaling
3.7.4. Loss aggregation
Also talk about replacement thresholds here.
3.8. Command-line support
Pelicun can be ran from the command line.
Installing the package enables the pelicun entry point, which points to tools/DL_calculation.py.
DL_calculation.py is a script that conducts a performance evaluation using command-line inputs.
Some of those inputs are paths to required input files, including a JSON file that provides most evaluation options.
3.8.1. Input file auto-population
It is possible for the JSON input file to be auto-populated (extended to include more entries) using either default or user-defined auto-population scripts.
3.9. Standalone tools
3.9.1. Unit conversion
3.9.2. Fit distribution to sample or percentiles
3.9.3. Random variable classes
3.10. Feature overview and examples
A series of examples, organized by feature, demonstrate the capabilities supported by pelicun.