Multinomial logistic regression

When to consider it

• Unordered categorical outcome with three or more levels.

Assumptions and scope

• Observations must be independent between subjects; repeated rows from one subject are not independent samples.
• Smallest outcome code is the reference. Inspect sparse categories, separation and independence of irrelevant alternatives.

Required inputs

• Unordered outcome, predictors — Numeric observations; preserve the row and column structure described above.

Review missing values and data coding before running. Sample inputs are synthetic teaching examples.

Implementation

Outcome first: 3–10 integer-coded unordered categories, then numeric predictors. Smallest observed outcome code is the reference. Independence of irrelevant alternatives is assumed, not tested. Explicit dummy coding required for categorical predictors. Synthetic example: 240 simulated three-category outcomes and two predictors.

• statsmodels 0.14.4: statsmodels.api.MNLogit

These are pinned library versions; saved results record the versions actually used at runtime.

Library references

• SciPy statistical methods
• statsmodels models and statistical tests
• scikit-learn clustering
• lifelines survival comparisons