Free AWS AIF-C01 Certified AI Practitioner Actual Exam Questions - Question 11 Discussion
well on the training dataset. When the company deployed the model to production, the model's
performance decreased significantly.
What should the company do to mitigate this problem?
This looks like a classic overfitting issue, so C makes the most sense to me.
It’s C because more diverse training data usually helps the model generalize better.
C/D? Adding more data (C) should help generalization, but if the model wasn’t trained long enough initially, giving it more training time (D) could also improve performance without overfitting.
C vs D but more training time without new data won’t fix overfitting.
C. Adding more data can help if the model is overfitting, but it’s also about the quality and representativeness of the data. If the training set is too narrow, more diverse data can improve real-world performance.
This looks like a classic overfitting case since it did well on training but poorly in production. Increasing training data (C) usually helps generalize better and avoid overfitting.
C. Adding more data usually helps because the model can learn a wider variety of patterns and generalize better. Overfitting often happens when the model only sees a limited dataset that doesn’t reflect real-world variability. Just increasing training time or tweaking hyperparameters might not solve the core problem if the data itself isn’t diverse enough. Reducing data volume (A) sounds counterproductive since less data often means less generalization. So, focusing on getting more representative data to train on seems like the best bet here.
Isn’t the main issue here that the model might be overfitting? Would adding more data (option C) always help, though? Could there be other reasons why performance dropped in production?