Bivariate Analysis — Scatter, Correlation, Crosstab — Data Analysis & Feature Engineering

Bivariate analysis examines the relationship between two variables at a time, the stage of EDA where the data begins to tell stories about how its features connect. It exists because the predictive power and insight in a dataset live in relationships — how one variable moves with another — and these connections are invisible to univariate analysis, which sees each variable alone. Without bivariate analysis, analysts miss the correlations that drive predictions, the associations that reveal data structure, and the confounding patterns that mislead naive conclusions. The right bivariate tool depends on the variable types involved: scatter plots and correlation for two continuous variables, box plots and group comparisons for continuous-versus-categorical, and crosstabs for two categoricals. Mastering this type-aware toolkit is what turns a collection of individually-understood variables into an understanding of how the data actually behaves.

Analogy🏏Cricket

🏏 Think of it like cricket: Reducing twenty batting statistics to two dimensions with PCA produces a flat projection that may look like one undifferentiated cloud, while t-SNE or UMAP finds that the data actually organises into distinct clusters — aggressive pinch-hitters, steady anchors, explosive finishers — whose separation PCA's flat projection smeared together. Just as the non-linear reduction reveals the natural groupings that flat projection could not, t-SNE and UMAP reveal structure that PCA misses because it can only flatten, not curve to follow the data's natural shape. The insight is that non-linear reduction methods follow the data's true curved geometry rather than forcing a flat projection, revealing the cluster structure and local neighbourhoods that linear methods like PCA cannot preserve.