Kirk Borne

❌If You’re Still Worshipping Pearson Correlation, You’re Not a Data Scientist — You’re Driving a Horse Cart in the Age of AI: https://valeman.medium.com/if-youre-still-worshipping-pearson-correlation-you-re-not-a-data-scientist-you-re-driving-a-831dc0590de6 ✨My summary: 💎 Don’t just plug your data into some formula to find the answers to the questions you started with. Real Data Science seeks and discovers the hidden questions in your data that you should be asking. Find the questions first, then seek their answers. 💎 Don’t just use your data. Allow your data to use you — to challenge your assumptions, to twist your thinking, to slap you into seeing the greater realities that the data represents, to shock and surprise you, and to take you down rabbit holes into a data wonderland you never imagined. Your data should point the way and guide you, not vice versa. 💎 One number is not a pattern. Find the meaningful patterns in the complex fabric of your data. Be sure you know (and be prepared to change) what you consider to be “meaningful”. Challenge your assumptions and your definitions as the deeper patterns in your data unfold. 💎 Look at your data. Master the art and science of data profiling — having your first date with your data. Remember Anscombe’s quartet and the Datasaurus Dozen: https://www.research.autodesk.com/publications/same-stats-different-graphs/ 💎 The natural data structure of the world is not rows and columns, but a graph. “All the world is a graph!" Shakespeare said “All the world’s a stage”, though he could just as well said “graph”, since both graphs and stage plays are all about entities and their relationships. And so is the world!🌐 💎 Outliers are not annoyances to dismiss. Three-sigma clipping might be throwing away your most significant, meaningful, valuable, brilliant discovery. Been there. Done that. Never again!🏆