CLI === Get help for the available (sub-)commands: ```shell ecsea --help ``` ## Training Train a new ECSEA model: ```shell ecsea train \ -tm MIN \ -ts 5000 \ -th 0.8 \ resources/test_data/test-low-level-log.xes \ resources/test_data/test-high-level-log.xes \ MyModel.model ``` Get help: ```shell ecsea train --help ``` ### Parameters Parameters that can be used in the training phase: | Parameter Key | Description | Valid Values | Default Value | |-------------------------------|----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|---------------------------------------------------------|---------------| | `train-test-split` | Defines the rate for the size of the training set (should be around 0.8) | 0.0 - 1.0 | 0.8 | | `time-span` | Defines the time-span in ms for the merging of "near" events. | \> 0 | 2000 | | `group-attributes` | Defines a list of event attributes which are used to assign events to the same window. | list of existing event attributes (e.g. 'org:resource') | [] | | `mapping-threshold` | Defines a mapping-threshold which is used in the low-level event merging step by comparing the calculated error score with the threshold. The low-level events are only merged if the error score is below this threshold. | \> 0.0 | 1.0 | | `timestamp-merge-type` | Defines the merge type of the timestamps of two or more events. | MIN,MAX,MEAN,MEDIAN | MEAN | | `mapping-weights-generations` | Defines how many generation the mappings weights should be learned (Basic genetic algorithm). If =0, no weights are learned. | \>= 0 | 0 | | `mapping-weights-population` | Defines the population size of one generation for the learning of mapping weights (Basic genetic algorithm). | \> 10 | 10 | ### Parameter Optimization Use the `-op` option in the CLI application to enable the hyperparameter optimization for the parameters `time-span`, `mapping-threshold`, and `timestamp-merge-type`. Enable the optimization for a given parameter `x` with the parameter `--optimize-x`. Look at `--help` for more information. ## Inspection of the model It is possible to get insights of a trained model. This command prints useful information: ```bash ecsea inspect-model --show-all MyModel.model ``` ## Update the model It is possible to update the parameter of a model (WIP!). Attention: Changing the parameters may influence the abstraction algorithm and that may result in unexpected behaviour! ```bash ecsea update-model -ga org:resource,spm:component myModel.model myNewModel.model ``` Options: - `-ga`: Change the mapping attributes - `-tm`: Change the timestamp merge type - `-ts`: Change the time span - `-th`: Change the mapping threshold ## Application Import a low-level log and a model and create a high-level log: ```shell ecsea apply \ resources/test_data/test-low-level-log.xes \ MyModel.model ``` Get CLI help: ```shell ecsea apply --help ```