Dimensionality Reduction
The short answer · 14 words
In short
Dimension reduction maps data with many features in fewer dimensions; information can be lost.
Flow-alerta.cab (2026): Dimensionality Reduction. https://flow-alerta.cab/en/glossar/dimensionality-reduction. Editorially reviewed on: 09 October 2026.
How much detail would you like?
Practical
Relevant for exploratory data analysis and visualization; selecting procedures by destination and data.
Possible example: A team presents a data set with many customer characteristics as a two-dimensional map and then checks how stable the visible patterns are.
Customer and product segmentation, noise filtering before classification models and visual analysis of operational data.
At a glance
- Core
- Linear methods such as PCA and non-linear methods such as t-SNE or UMAP produce low-dimensional representations. They can facilitate analysis and visualization. The benefit for a subsequent model must be verified; visible clusters are not an automatic proof of real groups.
- Use
- Relevant for exploratory data analysis and visualization; selecting procedures by destination and data.
- Example
- Possible example: A team presents a data set with many customer characteristics as a two-dimensional map and then checks how stable the visible patterns are.
- Limits
- Information and variance loss; misinterpretation of cluster distances at t-SNE (geometric distortions).
In context
Questions about this term
Is PCA or UMAP better?
That depends on the objective. PCA is linear; UMAP can represent non-linear structures. Both can distort or lose information.
Is t-SNE a general pre-processing for training?
It is primarily used for visualization; a suitability for downstream models is not automatically given.
All details and background
Linear methods such as PCA and non-linear methods such as t-SNE or UMAP produce low-dimensional representations. They can facilitate analysis and visualization. The benefit for a subsequent model must be verified; visible clusters are not an automatic proof of real groups.
ELI5 – Explained simply
Like the shadow cast of a 3D body on the wall: one dimension less, but the shape remains recognizable.
Why it matters in practice
Relevant for exploratory data analysis and visualization; selecting procedures by destination and data.
Example
Possible example: A team presents a data set with many customer characteristics as a two-dimensional map and then checks how stable the visible patterns are.
Pitfalls and limits
- Information and variance loss; misinterpretation of cluster distances at t-SNE (geometric distortions).
Origin & history
PCA belongs to the classic linear methods. t-SNE was described by van der Maaten and Hinton 2008; McInnes et al. published UMAP 2018.
Use in mid-sized companies
- Customer and product segmentation, noise filtering before classification models and visual analysis of operational data.
Related terms
- PCA
- t-SNE
- UMAP
- Feature Engineering
- Curse of Dimensionality
Matching Flow-alerta service
AI opportunity analysis & roadmapAI Potential Analysis & Roadmap – exploratory analysis as the basis of an AI projectMatching tool / use case
View use casesUse Case: Visualize data structures and compare the results with the original data set.Sources
- scikit-learn.orgscikit-learn, Decomposing signals in components
- arxiv.orgMcInnes et al. (2018), UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction
- www.jmlr.orgvan der Maaten & Hinton (2008), Visualizing Data using t-SNE
Quellen geprüft am 09.10.2026.
Editorially reviewed on: