AML Data Analysis & Reporting 4

Question 1

Which technique is used to reduce dimensionality while preserving the maximum variance in the data?

Accepted Answer

Principal Component Analysis (PCA)

Answer

PCA projects data onto orthogonal components ordered by explained variance, maximally preserving variance in fewer dimensions.

Question 2

A model's training loss continues to decrease while validation loss increases after epoch 20. What does this report?

Accepted Answer

Overfitting

Answer

Divergence between training and validation loss curves is the classic signature of overfitting — the model memorizes training data but fails to generalize.

Question 3

In A/B testing analysis, what is the risk of running multiple comparisons without correction?

Accepted Answer

Inflated Type I error rate (false positives)

Answer

Each additional comparison increases the probability of at least one false positive; Bonferroni or FDR corrections control this inflated error rate.

Question 4

What does SHAP (SHapley Additive exPlanations) provide in ML reporting?

Accepted Answer

Per-prediction, feature-level attributions explaining individual model outputs

Answer

SHAP values assign each feature a contribution to a specific prediction, enabling both local (per-instance) and global interpretability.

Question 5

A report shows Mean Absolute Error (MAE) = 5 and Root Mean Squared Error (RMSE) = 15 for a regression model. What does this gap suggest?

Accepted Answer

There are significant outlier errors driving up RMSE

Answer

RMSE penalizes large errors more than MAE; a large MAE-RMSE gap signals that some predictions have very large errors (outliers in residuals).

(AML) Advanced Machine Learning Certification Practice Test

(AML) Advanced Machine Learning Certification Practice Test

AML Data Analysis & Reporting 4