Understanding Your Data: Exploratory Data Analysis

Two classes can have exactly the same average mark and still be completely different classrooms. The average does not lie, but it does not tell you everything. This video shows how to look at a table of data before any model is trained: the first questions to ask, how to read one column, and how to read two columns together. One small made-up table of students is used from start to finish. You will learn: The first questions: how big is the table, what type is each column, what is missing, what is repeated Mean against median, and why one unusual value moves the mean much more Spread: the range and the standard deviation (population version, divide by n, and how the sample version differs) A histogram built bar by bar, and a box plot built from the five-number summary A scatter plot and the correlation coefficient, and why correlation is not causation Three small datasets with the same summary numbers and three different shapes What to write down before you start modelling Chapters 0:00 Same average, two different classes 0:39 First questions: size and types 1:31 Missing values and repeated rows 2:08 One column: mean and median 3:24 Spread: range and standard deviation 4:09 A histogram, bar by bar 4:53 A box plot from five numbers 5:37 Two columns: scatter plot and correlation 6:25 Correlation is not causation 6:56 Same numbers, different shapes 7:29 What to write down before modelling 8:01 Recap + try it yourself Try it yourself: take the numbers 3, 5, 6, 8 and 28. Find the mean and the median. Remove the 28 and find them again. Which one moved more? Animations made with Manim Community Edition. The narration is a synthetic voice. #DataScience #MachineLearning #AIEngineer More from Harban AI Labs: https://harban-ai-labs.vercel.app/