Biplot¶

A joint low-dimensional display of observations and variables from a data matrix, typically overlaying row scores with column loadings derived from a matrix factorization.

Version: v2 · 2026-09-19 · History
Domain-specific #: 3479
Domain group: Formal Sciences
Origin domain: Experimental Design & Statistics
Subdomain: Multivariate Statistics and Visualization → Experimental Design & Statistics

Core Idea¶

Scaling choices determine whether distances, angles, projections or inner products are interpretable; PCA, correspondence and generalized biplots use different metrics and representations. A centered, scaled or transformed matrix is approximated by low-rank factors, singular values are allocated between row and column coordinates, and both sets are plotted in one plane with a declared reconstruction rule. The abstraction is therefore identified by a declared carrier, a transformation or constraint over that carrier, and an invariant that tells an analyst whether the named structure is genuinely present.

How would you explain it like I'm…

Dots and Arrows Picture

Imagine a big chart about a class: each kid's height, speed, and how many books they read. A Biplot squishes all that onto one flat picture, with a dot for each kid and an arrow for each thing you measured. It's a helpful sketch, but squishing means some details get lost, and you need to know how it was drawn to read it right.

A Whole Table in One Picture

A Biplot is a picture that shows a big table of numbers in one flat drawing. Each row of the table, like a person, gets a dot, and each column, like a test score, gets an arrow or marker, drawn on the same picture. It is made by boiling the table down to its two most important directions, so the picture is a simplified version of the whole table. If you slide a dot straight onto an arrow's line, where it lands gives an estimate of that row's value for that column. How you are allowed to read other things, like distances between dots, depends on choices made when building the plot.

Joint Plot of Rows and Columns

A Biplot displays both the observations (rows) and the variables (columns) of a data table in a single two-dimensional plot. First the data is prepared, for example by centering and scaling, and then approximated by a low-rank factorization, usually through the singular value decomposition, keeping only the top two dimensions. The row points and column markers are chosen so that combining a row with a column, by projecting the point onto the variable's direction, approximately reconstructs that table entry. How the singular values are split between rows and columns is a scaling choice, and it decides whether distances between observations, or angles between variables, can be interpreted meaningfully. Related versions, such as PCA biplots and correspondence analysis biplots, use different distance measures, so you need to know which kind you are reading.

A Biplot represents a data matrix by plotting row coordinates (observations) and column coordinates (variables) together in one low-dimensional space, typically a plane. The matrix is first centered, scaled, or otherwise transformed, and then approximated by low-rank factors, commonly via the singular value decomposition X ≈ U_k D_k V_k^T with k = 2. The singular values are allocated between row and column coordinates, often parameterized by a scaling exponent alpha, giving row scores U D^alpha and column markers V D^(1-alpha), so that inner products between a row point and a column marker approximate the corresponding entries under the declared reconstruction rule. The choice of alpha determines which other geometric features are interpretable: one scaling preserves distances between observations, another makes angles between variable vectors reflect their correlations. PCA biplots, correspondence-analysis biplots, and generalized biplots differ in the metric used and in how categorical data are represented. A properly reported biplot states the preprocessing, missing-data handling, rank, metric, scaling, variance explained by the axes, and the limits of interpretation.

Scope of Application¶

Biplot belongs to multivariate statistics and visualization and is useful where the analyst can specify the typed multivariate statistics and visualization carrier, defining objects and relations, parameters, conventions, evidence, boundary cases, and comparison targets, then evaluate the observations and variables, data types, centering and scaling, missing-data handling, matrix factorization and rank, metric, singular-value allocation or alpha scaling, row scores and column markers, axes and variance explained, categorical treatment, uncertainty and interpretation limits are explicit.

Clarity¶

The abstraction clarifies a crowded vocabulary by making the observations and variables, data types, centering and scaling, missing-data handling, matrix factorization and rank, metric, singular-value allocation or alpha scaling, row scores and column markers, axes and variance explained, categorical treatment, uncertainty and interpretation limits are explicit the center of the account. A claim should name the carrier, the governing operation or relation, the applicable assumptions, and the recognition test.

Manages Complexity¶

Without the abstraction, an analyst must reason directly over many local details: the carrier roles, admissibility assumptions, competing conventions, derived invariants, boundary cases, and proof or validation obligations specific to Biplot. Biplot compresses them into the roles in the structural signature. That compression permits comparison across instances without erasing the variables that determine validity. It also exposes which details may be varied safely and which are constitutive.

Abstract Reasoning¶

Identify the carrier. State what the elements, states, objects, or observations are: the typed multivariate statistics and visualization carrier, defining objects and relations, parameters, conventions, evidence, boundary cases, and comparison targets. Reject examples whose alleged carrier belongs to a different problem. 2.

Knowledge Transfer¶

Knowledge transfers strongly among subfields of multivariate statistics and visualization because they reuse the typed multivariate statistics and visualization carrier, defining objects and relations, parameters, conventions, evidence, boundary cases, and comparison targets, A centered, scaled or transformed matrix is approximated by low-rank factors, singular values are allocated between row and column coordinates, and both sets are plotted in one plane with a declared reconstruction rule., and type the carrier, state every parameter and convention in the definition, test that the observations and variables, data types, centering and scaling, missing-data handling, matrix factorization and rank, metric, singular-value allocation or alpha scaling, row scores and column markers, axes and variance explained, categorical treatment, uncertainty and interpretation limits are explicit, compare the nearest accepted identity, and report counterexamples, uncertainty, and limiting cases.

Relationships to Other Abstractions¶

Current abstraction Biplot Domain-specific

Parents (1) — outgoing typed relations

Biplot is a kind of Representation Prime

It remains its own entry because its identity is fixed by the observations and variables, data types, centering and scaling, missing-data handling, matrix factorization and rank, metric, singular-value allocation or alpha scaling, row scores and column markers, axes and variance explained, categorical treatment, uncertainty and interpretation limits.

Hierarchy path (1) — routes to 1 parentless root

Biplot → is a kind of Representation → presupposes Abstraction

Neighborhood in Abstraction Space¶

Biplot sits in a crowded region of the domain-specific corpus (19^th percentile for distinctiveness): several abstractions share nearly its structure, so a description that fits it tends to fit its neighbors too.

Family — Data Visualization Chart Forms (10 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-25