When is a Biased Estimator Better? A Look at Ratio Estimators

Efficient NLP · Intermediate ·🛡️ AI Safety & Ethics ·1y ago
Try Voice Writer - speak your thoughts and let AI handle the grammar: https://voicewriter.io The ratio estimator is an example where a biased estimator is better than an unbiased one. Many things are ratios, eg, the view-to-like ratio for Youtube videos. While the bias can be corrected, there are good reasons not to do so, including the bias-variance tradeoff. I also demo a numerical simulation using real data from Lex Fridman's channel. 0:00 - Bias of an estimator 0:47 - Sample variance bias correction 2:32 - Ratio estimator and example 4:01 - Bias correction methods for ratio estimator 5:0…
Watch on YouTube ↗ (saves to browser)

Chapters (7)

Bias of an estimator
0:47 Sample variance bias correction
2:32 Ratio estimator and example
4:01 Bias correction methods for ratio estimator
5:05 Why not correct the bias
6:07 Bias-variance tradeoff
7:32 Numerical simulation
AI isn’t just for builders… it’s for everyone.
Next Up
AI isn’t just for builders… it’s for everyone.
Coursera