Exam AWS Certified Machine Learning - Specialty topic 1 question 152 discussion

Dimensions are too high. Use PCA

A should be the answer to avoid the curse of dimensionality

Easy choice. Always choose PCA for dim reduction

the best feature engineering strategy for the ML specialist to use with Amazon SageMaker is to apply dimensionality reduction by using the PCA algorithm.

Selected Answer: A Given that the dataset has 1,020 features and 200 of them are highly correlated, it is likely that the dataset suffers from multicollinearity. In such cases, dimensionality reduction techniques like principal component analysis (PCA) can be used to transform the data into a lower dimensional space without losing much information. Therefore, option A, "Apply dimensionality reduction by using the principal component analysis (PCA) algorithm" is the most appropriate feature engineering strategy for the ML specialist to use with Amazon SageMaker. This would help reduce the computational complexity of the model, improve model performance, and help to avoid overfitting.

A. Apply dimensionality reduction by using the principal component analysis (PCA) algorithm. Since the dataset has many features, and a significant number of them have high correlation scores, the model may suffer from the curse of dimensionality. To reduce the dimensionality of the dataset, the specialist can use a technique like PCA, which reduces the number of features while still retaining the maximum amount of information. PCA can help remove redundant features and improve the model's performance by reducing the chances of overfitting. Additionally, since there are no missing values and a small percentage of duplicate rows, no data cleaning techniques like anomaly detection or dropping the features are required. Concatenating features with high correlation scores is not an appropriate strategy since it may lead to collinearity issues.

A PCA: PCA is a linear dimensionality reduction technique (algorithm) that transforms a set of correlated variables (p) into a smaller k (k<p) number of uncorrelated variables called principal components while retaining as much of the variation in the original dataset as possible

Choosing C is answer by ExamTopics is completely laughable.

I think it's A.