Logistic regression can be bionomial, _______ or multinomial.

Question

Accepted Answer

The correct option is (1):
The correct option is **ordinal**. Logistic regression is a core classification algorithm whose variations are primarily defined by the scale and number of categories in the dependent (target) variable. The three main forms of logistic regression are indeed listed in this question: **Binomial**, **Ordinal**, and **Multinomial**. Binomial Logistic Regression, also widely known as Binary Logistic Regression, is used when the target variable has exactly two possible outcomes (e.g., Yes/No, Pass/Fail). When the target variable has three or more categories, we must distinguish between two types: Ordinal Logistic Regression is specifically applied when the multiple categories have a natural, meaningful rank or order (e.g., Low/Medium/High, Strongly Disagree/Disagree/Neutral/Agree/Strongly Agree).

The incorrect option is (2):
**Numbered** is not a standard, formal classification for a type of logistic regression model. In statistics and machine learning, an outcome variable that is simply 'numbered' could be interpreted in several ways. If the number represents a continuous measurement (like temperature or height), then standard linear regression would be more appropriate, not logistic regression. If the numbers represent distinct, fixed categories (like 1, 2, 3), the outcome is essentially categorical, meaning the appropriate model would be Binomial, Ordinal, or Multinomial, depending on whether the numbers are ordered and how many categories exist. Thus, 'numbered' lacks the precise conceptual meaning required to define a model type.

The incorrect option is (3):
The term **unimial** is not a recognized classification in the context of general linear models (GLMs) or logistic regression. The standard terms used to describe the number of outcome classes are *bi*-nomial (for two classes) and *multi*-nomial (for three or more classes). The prefix 'uni-' implies a single outcome, which is statistically trivial for a prediction model. All predictive models, including logistic regression, inherently require at least two possible outcomes to model a choice or probability.

The incorrect option is (4):
The option **binary** is incorrect because it is synonymous with **binomial** logistic regression, which is already mentioned in the question: "Logistic regression can be bionomial, \_\_\_\_\_\_\_ or multinomial." Using 'binary' as the missing term would make the statement redundant ("Logistic regression can be binomial, binary, or multinomial") and would still leave out the distinct and essential third category: Ordinal. Binary and Binomial regression models are structurally identical, both used for two-class classification problems.