monomodal

Monomodal is an adjective used primarily in statistics and data analysis to describe a distribution, data set, or phenomenon that has a single mode. The mode is the value or values at which observations are most frequent, or, for a probability density, the point where the density attains a local maximum. A monomodal distribution has one dominant peak.

While monomodal is often used interchangeably with unimodal, some writers prefer unimodal to emphasize a single peak in the distribution, whereas monomodal stresses the presence of one mode. In formal terms, a distribution is monomodal if it has exactly one local maximum; discrete distributions may have a unique value with the largest probability mass.

Examples: The normal distribution is a classic monomodal/unimodal distribution with a single peak at its mean. The exponential distribution has a single mode at zero. A mixture of two normals is bimodal or multimodal, not monomodal. Some unimodal distributions may be asymmetric or skewed.

Broader usage: In data science and machine learning, monomodal can also describe datasets that contain a single data modality (for example, text-only data) as opposed to multimodal datasets that combine multiple modalities (text, image, audio).

In such contexts, monomodal emphasizes the absence of multiple distinct data types.

See also: unimodal, multimodal, mode, probability distribution.