Exam AWS Certified Machine Learning - Specialty topic 1 question 254 discussion

BCE B. Decrease the amount of regularization that the model uses: Regularization is used to prevent overfitting, but if the fitted model has poor accuracy on both the training and test datasets, reducing the amount of regularization can help the model better capture the underlying patterns and improve its accuracy. C. Increase the number of training examples that the model uses: Increasing the number of training examples allows the model to learn from a larger and more diverse dataset, which can help improve its ability to generalize and make accurate predictions. E. Increase the number of model features that the model uses: Adding more relevant features to the model can enhance its ability to capture important patterns and relationships in the data, leading to improved accuracy.

since model is underfitting, reduce the regularization will allow to use the features more, large no. of training example meaning more learning. and increase features will help model understand the establish pattens better

Please someone corrects me if I am wrong but I don't see that the question mentions overfitting or underfitting. It tells that both training and test datasets have poor accuracy. For this reason, I wouldn't apply B and I would go with the general steps that would help me to improve model accuracy (CDE)

B and C obvious. Between E and F , I would start by increasing features before considering reduction

I will go with BCE, other options are for solving overfitting

A. NO - regularization will reduce overfitting, not accuracy B. YES - to much regularization will reduce complexity and thus decrease accuracy C. YES - the more data the merrier D. NO - test examples will no influence model performance E. YES - the more features the more there is to learn F. NO - as per E

B C and E

We have an underfitting problem here. To remedy this, we must follow the alternatives that increase the complexity of the model: Letter B - C - E.

@RRST summarized well.

We can decrease the overfitting by reducing the number of features, so how is F not an answer

The problem is stating the Underfitting scenario. So correct answers are ACE

A. Increase the amount of regularization that the model uses. C. Increase the number of training examples that the model uses. E. Increase the number of model features that the model uses.