How to simplify complex data by finding the hidden axes of variation
In short: Principal component analysis is a powerful mathematical technique that transforms messy, correlated datasets into a simpler, independent coordinate system. By identifying the directions of greatest variance, it allows researchers to visualize complex information, reduce dimensionality, and uncover the underlying structure of data across fields from genetics to atmospheric science.
At its core, principal component analysis (PCA) is a linear dimensionality reduction method. It functions by rotating a dataset into a new coordinate system where the axes, known as principal components, represent the directions of maximum variance. The first component captures the largest possible variance, while each subsequent component accounts for the remaining variance while remaining orthogonal to the previous ones. This process effectively creates an orthonormal basis where individual dimensions are linearly uncorrelated.
Mathematically, PCA identifies these components as the eigenvectors of the data's covariance matrix. To perform the analysis, one must first center the data by subtracting the mean of each variable. Once the covariance matrix is computed, eigendecomposition or singular value decomposition is used to extract the eigenvectors. These vectors are then normalized to unit length, serving as the axes of an ellipsoid fitted to the data. The proportion of total variance explained by each component is determined by dividing its corresponding eigenvalue by the sum of all eigenvalues.
Invented by Karl Pearson in 1901 and later developed by Harold Hotelling in the 1930s, PCA is known by many names depending on the discipline, including the Karhunen–Loève transform in signal processing and empirical orthogonal functions in meteorology. While it is the simplest of the eigenvector-based multivariate analyses, it is distinct from factor analysis, which incorporates more domain-specific structural assumptions. Today, researchers use tools like biplots and scree plots to interpret these components, enabling them to identify clusters and simplify datasets that contain many highly correlated variables.
Opens the Intellect+ feed with this card first. Swipe up for the next one.
Sources
- Summary by Intellect · source: Principal component analysis
To cite this card, link to https://intellectplus.co.uk/app/technology-principal-component-analysis.
How we source, check and correct cards: Editorial standards.
General information, not professional advice. Spotted an error? Flag the card or see corrections and complaints.
Intellect Plus is a visual educational discovery platform for exploring fascinating facts and ideas. Brain Food, Not Brain Rot. What is Intellect Plus?
Related
More in Technology today · More in Technology · Latest · A–Z · All topics