MLS-C01 · Question #178
A machine learning (ML) specialist must develop a classification model for a financial services company. A domain expert provides the dataset, which is tabular with 10,000 rows and 1,020 features. Dur
The correct answer is A. Apply dimensionality reduction by using the principal component analysis (PCA) algorithm.. Given a high-dimensional dataset (1,020 features) with many highly correlated feature pairs, applying Principal Component Analysis (PCA) is the most suitable feature engineering strategy. PCA effectively reduces dimensionality by transforming correlated features into a smaller se
Question
A machine learning (ML) specialist must develop a classification model for a financial services company. A domain expert provides the dataset, which is tabular with 10,000 rows and 1,020 features. During exploratory data analysis, the specialist finds no missing values and a small percentage of duplicate rows. There are correlation scores of > 0.9 for 200 feature pairs. The mean value of each feature is similar to its 50th percentile. Which feature engineering strategy should the ML specialist use with Amazon SageMaker?
Options
- AApply dimensionality reduction by using the principal component analysis (PCA) algorithm.
- BDrop the features with low correlation scores by using a Jupyter notebook.
- CApply anomaly detection by using the Random Cut Forest (RCF) algorithm.
- DConcatenate the features with high correlation scores by using a Jupyter notebook.
How the community answered
(20 responses)- A80% (16)
- B5% (1)
- C5% (1)
- D10% (2)
Why each option
Given a high-dimensional dataset (1,020 features) with many highly correlated feature pairs, applying Principal Component Analysis (PCA) is the most suitable feature engineering strategy. PCA effectively reduces dimensionality by transforming correlated features into a smaller set of uncorrelated components, addressing multicollinearity and improving model efficiency.
Given a high-dimensional dataset with 1,020 features and 200 pairs exhibiting high correlation (> 0.9), applying Principal Component Analysis (PCA) is an appropriate feature engineering strategy. PCA effectively reduces dimensionality by transforming correlated features into a smaller set of orthogonal components, thereby mitigating multicollinearity and potentially improving model performance and reducing training time for a classification model.
Dropping features with low correlation scores is a feature selection method that may not effectively address the problem of high correlation between features (multicollinearity) when dealing with a very large number of features.
Random Cut Forest (RCF) is an algorithm specifically designed for anomaly detection, which is not the primary goal here; the specialist needs to prepare features for a classification model.
Concatenating features with high correlation scores would not resolve multicollinearity but rather create new features that are also highly correlated, potentially worsening the problem for the classification model.
Concept tested: Dimensionality reduction for multicollinearity
Source: https://docs.aws.amazon.com/sagemaker/latest/dg/pca.html
Topics
Community Discussion
No community discussion yet for this question.