Which methods can the Data Scientist use to improve the model performance and satisfy the Marketing team’s needs?
(Choose two.)
Add L1 regularization to the classifier
Add features to the dataset
Perform recursive feature elimination
Perform t-distributed stochastic neighbor embedding (t-SNE)
Perform linear discriminant analysis
Explanations:
Adding L1 regularization (Lasso) helps reduce overfitting by penalizing irrelevant features, thus improving generalization and reducing the gap between training and validation accuracy.
Adding more features can worsen overfitting, especially in the absence of insights from the Marketing team. It may increase complexity without improving model performance.
Recursive feature elimination (RFE) helps identify and remove irrelevant or redundant features, improving model performance by selecting only the most relevant features.
t-SNE is a dimensionality reduction technique used for visualization. It does not directly improve model performance or address issues like overfitting or feature relevance.
Linear Discriminant Analysis (LDA) is a dimensionality reduction technique used for classification but is not specifically designed to address overfitting or feature selection in logistic regression.