Why is changing a scale parameter not equivalent to changing a distribution's shape?
StatisticsLevel: IntermediateLikelihood and fitted distributions
Why is changing a scale parameter not equivalent to changing a distribution's shape?
Hints
- Location–scale transforms preserve standardized shape.
- Tail exponents can be shape parameters.
Answer
A positive scale stretches standardized values uniformly; shape parameters can change skewness or tail behavior. Check parameterization before comparing fitted models.
Source and licence: SciPy — distribution fitting and likelihood · BSD-3-Clause · answer: Quantoros editorial