# Principal component analysis (PCA) as a feature compressor

Source: https://onco.cc/terms/pca/  
OnCo record `pca` (Term). Data CC BY-NC 4.0, attribute "Data from OnCo (onco.cc)"; commercial use needs a licence.

## TL;DR

PCA rotates the data onto the directions of greatest variance and keeps the top few, compressing twenty thousand genes into a few hundred numbers before a model sees them.

## Summary

Principal component analysis is a linear dimensionality reduction technique used in exploratory analysis and preprocessing (Wikipedia). Reducing a full transcriptome to about 256 components before ridge or Cox regression is a strong, cheap baseline and the extraction recipe for frozen foundation-model embeddings; reducing each modality separately before concatenation stops a small block (a few hundred proteins) being swamped by a large one (all genes). The components must be fitted on the training fold only.

## Fields

- Kind: Term
- Last checked: 2026-09-24
- Also known as: principal component analysis; PCA; principal components; PCA-256; PCA reduction; separate-block PCA; block-wise PCA; dimensionality reduction
- Tags: cansim-terms

## Notes

- Listed in the CanSim terms map 1.0.0 (docs/onco/terms.json, generated 2026-09-24), CC BY 4.0, attribution: CanSim project, an open, public-data-first cancer foundation-model programme; CanSim page path /terms/pca.

## Sources

- Wikipedia: https://en.wikipedia.org/wiki/Principal_component_analysis
- Wikipedia: https://en.wikipedia.org/wiki/Principal_component_analysis

## Connected records

- terms: [Cancer AI vocabulary (CanSim terms map)](https://onco.cc/terms/cancer-ai-vocabulary/), [Data leakage in model evaluation](https://onco.cc/terms/data-leakage/), [Embedding (learned representation)](https://onco.cc/terms/embedding/), [Multimodal fusion (early, late, modality dropout)](https://onco.cc/terms/multimodal-fusion/), [Out-of-distribution detection (Mahalanobis guard)](https://onco.cc/terms/ood-detection/), [Ridge regression and the multilayer perceptron](https://onco.cc/terms/ridge-regression/), [Spatially aware clustering (SpaGCN, KNN smoothing)](https://onco.cc/terms/spagcn/)

---
JSON: https://onco.cc/api/v1/entities/pca.json