Tutorvo
Menu
Courses Guides Pricing Log in Sign up

The Accuracy Trap: 3 Faults in 100 Runs, Confusion Counts & the «Always OK» Model

Lesson 1 of 815 minLab

All lessons
  1. The Accuracy Trap: 3 Faults in 100 Runs, Confusion Counts & the «Always OK» ModelLab15 minFree
  2. Metrics That Respect the Minority: Balanced Accuracy, the Matthews Coefficient, F1 & FβLab16 min Locked
  3. Random Undersampling & Oversampling: Given Draw Tables, New Class Shares & What Gets LostLab16 min Locked
  4. Synthetic Minority Samples: Neighbour Distances, the k Nearest & New Points by InterpolationLab17 min Locked
  5. Class Weights: n ÷ (k × n_class), a Weighted Log-Loss & a New Best Cut-OffLab16 min Locked
  6. Cost-Based Thresholds: A Cost Matrix, Expected Cost per Run & the Cheapest Cut-OffLab16 min Locked
  7. Probability Correction After Resampling: Training vs Real Prevalence & a Calibration TableLab16 min Locked
  8. Capstone – Comparing Strategies: Balanced Accuracy, MCC, Recall, Expected Cost & a Ranked SummaryLab18 min Locked

You are not signed in — the lab works, but your progress is not saved. Sign up

Loading the spreadsheet…
Next lesson →