The EDA Workflow and Checklist — Data Analysis & Feature Engineering

Exploratory data analysis is the disciplined first encounter between an analyst and a dataset, the stage where you build an intuition for what the data contains before committing to any model or conclusion. It exists because raw data arrives as an opaque grid of numbers and labels that hides its structure, its flaws, and its stories, and leaping straight to modelling on data you do not understand is the single most common cause of wasted effort and wrong conclusions. A systematic EDA workflow — a repeatable checklist of questions and checks — exists to replace ad-hoc poking around with a thorough, reproducible investigation that surfaces missing values, outliers, distributions, and relationships before they sabotage downstream work. Mastering this workflow is what separates analysts who trust their data from those who are repeatedly blindsided by it.

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.