Data, Representations, and Simple Prediction Problems
Due September 29, 2026 at 3:59 AM
Works through the fundamental skills of data representation, error measurement, constant prediction rules, and related topics.
Parts
Task 1: Data and Representations
Load the frozen flight CSV, inspect raw row dictionaries, extract columns, convert arrival delays into lists and arrays, and compare tabular representations.
Task 2: Losses and Constant Predictors
Implement MAE and MSE, compare constant guesses, search over scalar predictions, and explain why mean and median answer different loss functions.
Task 3: Matrices and Covariance
Build delay matrices, compute summary statistics and covariance, and interpret how flight-delay variables move together.
Task 4: Media Transformations
Recover unknown transformation strengths for audio fades and image edits using optimization, loss curves, and visual/audio diagnostics.
Task 5: Final Report
Connect data representation, model class, optimization, and evaluation in a concise written report.