PCA

PCA — Principal Component Analysis turns a wide table of correlated variables into a handful of components that capture most of the story. Type your data and instantly get the eigenvalues, the scree plot, the variance explained, the loadings and a biplot — so you can see which variables move together and how many dimensions you really need. Fully offline. Measure height, weight, waist, chest and hip, and you'll find they mostly move together — they're really one underlying thing, "size". PCA finds those hidden directions automatically: it rotates your variables into new axes (principal components) ordered by how much variation each one captures, so the first few components often summarise almost everything. No sign-up. No internet needed. No data collected. Just clear, instant statistics. WHAT YOU CAN DO • Data — type your dataset into a clean grid, one column per variable and one row per observation. Everything recomputes as you type; rows with blanks are skipped. • Scree — see each component's eigenvalue largest-first, with the Kaiser line at 1, so you can keep the components above it or stop at the "elbow" where the drop levels off. A biplot places observations on the first two components with the variables drawn as arrows. • Concept — learn what PCA does, what eigenvalues and loadings mean, and how variance is shared across components — with offline, beautifully typeset formulas that never overlap. • Simulate — build correlated data from a few hidden factors, set the correlation strength, number of variables, factors and observations with + / - steppered sliders, and watch the scree plot show exactly how many components survive. • Examples — load ready-made datasets (body measurements, cars and more) with one tap. WHAT YOU GET • Eigenvalues and the percentage of variance each component explains. • A scree plot with the Kaiser (eigenvalue > 1) line. • Component loadings — how each original variable feeds each component. • A biplot of observations and variable arrows. WHY YOU'LL LIKE IT • See the dimensions — the scree plot shows how few components you really need. • Read the loadings — find which variables share a component and point the same way. • Fully offline — every eigenvalue, plot and formula works with no connection. • Hands-on — type exact data, or build correlated data with sliders and + / - steppers. • Clean "Deep Turquoise" design, easy on the eyes. • Privacy-friendly — your data stays on your device. Whether you are a student meeting eigenvalues for the first time, a data analyst reducing dimensions before modelling, or anyone trying to make sense of many correlated measurements, PCA turns a wide table into a clear, compact picture in seconds. Download it, paste your variables, and see how few dimensions you really need.