Python API

While the CLI is convenient for ad-hoc runs, real integrations often use the Python API directly.

---

crml.loader

from crml.loader import load_crml

model = load_crml("model.yaml")

• Accepts YAML or JSON.
• Returns a Python dict.

---

crml.validator

from crml.validator import validate_crml

validate_crml(model)

• Raises jsonschema.ValidationError if the model does not conform to crml_schema.json.

---

crml.runtime

from crml.runtime import run_model, explain_model

results = run_model(model, runs=20000)
print(results)

print(explain_model(model))

run_model returns a dictionary:

{
  "EAL": 2.7136410e+07,
  "VaR_95": 3.9736940e+07,
  "VaR_99": 4.4349590e+07,
  "VaR_999": 6.8422817e+07
}

---

crml.frequency, crml.severity, crml.entropy

You can also call the low-level functions directly:

from crml.frequency import sample_gamma_poisson_frequency
from crml.severity import sample_mixture
from crml.entropy import entropy_from_counts

These are useful for:

• custom experimental models
• notebooks
• research papers