Syllabus Map

• Study map: Syllabus Study Map

Core Idea

• Bernoulli models binary outcomes.
• Binomial models count of successes.
• Gaussian models continuous noise and aggregate effects.

Practical Notes

Match distribution assumptions to data type.

• Wrong assumptions degrade calibration and fit quality.

Why This Matters for ML

• Data-type-specific modeling assumptions map directly to distribution choices.
• Bernoulli/Binomial link to binary/count tasks; Gaussian assumptions appear in noise models.
• Losses like MSE and logistic loss have probabilistic interpretations via these distributions.
• Calibration and uncertainty interpretation depend on correct distributional modeling.