A greedy approach called "forward selection" is used in statistics and machine learning to choose features. Finding a collection of features that most significantly improves a model's prediction accuracy is the aim. With this approach, the model is initially feature$-$free and the most important feature that enhances the model's performance is added iteratively. Accuracy is used as the assessment metric when applying a forward selection method to a classification task. After introducing a new feature, accuracy rises from $85\%$ to $90\%.$ What does this imply regarding the new feature? a$.$ The added feature improved the model's generalization. b$.$ The added feature introduced noise. C$.$ The added feature is irrelevant to the model. d$.$ The added feature increased model complexity unnecessarily.