Minimum Description Length
Definition
The MDL principle favors a model with a short joint description of model and data.
A model selection principle that favors a short joint description of model and data.
Simply explained
How to choose an explanation that is as concise as possible including its exceptions; counting only the rules is not enough.
Why it matters in practice
Helps realistically assess model quality, limits, and technical decisions.
Example from business
Between two models, the one is chosen that explains structure without introducing unnecessary complexity.
Limits and pitfalls
- Equating MDL with mere parameter count or omitting coding and model class.
Origin & history
Rissanen formulated MDL in the information theory of the 1970s.
Use in mid-sized companies
- Possible application: A data team compares suitable model classes with documented coding of model and data. It checks whether the choice depends on coding assumptions.
Frequently asked questions
What does the principle of minimum description length mean?
A model selection principle that favors a short joint description of model and data.
Which mistakes should be avoided?
Equating MDL with mere parameter count or omitting coding and model class.
Sources
Sources checked on 12.10.2026. Editorial application examples are hypothetical; no evidence of real customer results.
Editorially reviewed on: