Exam AWS Certified Machine Learning - Specialty topic 1 question 159 discussion

i will go for B, D and E. B and D for me are like doing partial regression and corr plot can actually tell you briefly how well the univariate is correlated with your target and i guess that also apply for D.. and E , feature importance ranking that's what feature selection strategy want from my POV. And for Data Binning is data enrichment just like augmentations , but then the question was saying they want to do feature selection over 1k+ variables which implies they actually care more about which variable(s) can contribute more on determining the price ?

B. Correlation plot with heat maps: This technique can be used to identify the relationship between each feature and the target variable (sales price). By creating a correlation plot with heat maps, the company can quickly visualize the strength and direction of the relationship between each feature and the target variable. D. Univariate selection: This technique can be used to select the features that have the strongest relationship with the target variable. It involves analyzing each feature independently and selecting the ones that have the highest correlation with the target variable. E. Feature importance with a tree-based classifier: This technique can be used to determine the most important features that contribute to the target variable. By using a tree-based classifier such as Random Forest or Gradient Boosting, the company can rank the importance of each feature and select the ones that have the highest importance.

For feature selection in machine learning, you can use the following techniques: B. Correlation plot with heat maps: Correlation analysis helps identify relationships between features and the target variable. A heat map can visually represent the correlation matrix, helping to identify highly correlated features. D. Univariate selection: Univariate selection methods evaluate the relationship between each feature and the target variable independently. Common techniques include statistical tests such as chi-squared tests, ANOVA, or mutual information. E. Feature importance with a tree-based classifier: Tree-based classifiers, such as decision trees or random forests, can provide feature importance scores. These scores help identify which features contribute the most to the predictive performance of the model.

A, C and F are not feature selection techniques.

BDE seem to be the only viable feature selection methods here

Those are the only ones for FS.

the most appropriate feature selection techniques for the company to determine the primary features contributing to the sales price are B (correlation plot with heat maps), D (univariate selection), and E (feature importance with a tree-based classifier).

BDE as stated here: https://towardsdatascience.com/feature-selection-techniques-in-machine-learning-with-python-f24e7da3f36e

BDE for me

throwing my weight behind B D E. Correlation with heatmaps help us eliminate multicollinearity, Univariate testing helps us see which ones are correlated with the target, same as feature importances of tree-based algorithms.

CDF for me