Exam AWS Certified Machine Learning - Specialty topic 1 question 26 discussion

This is a very tricky question. The idea is to reconfigure the ranges of the hyperparameters. A refers to a feature, not a hyperparameter. A is out. C refers to training the model, not optimizing the range of hyperparameters. C is out. Now it gets tricky. D will let you find determine what the approximately best tree depth is. That's good. That's what you're trying to do but it's only one of many hyperparameters. It's the best choice so far. B is tricky. t-SNE does help you visualize multidimensional data but option B refers to input variables, not hyperparameters. For this very tricky question, I would do with D. It's the only one that accomplishes the task of limiting the range of a hyperparameter, even if it is only one of them.

B doesn't make sense I think it's D

The goal is to reduce training time and costs by optimizing the hyperparameter tuning process. In tree-based ensemble models (e.g., XGBoost, Random Forest, or Gradient Boosting), tree depth is one of the most influential hyperparameters affecting: Model complexity: Deeper trees increase complexity but may lead to overfitting. Training time: More depth means more splits, significantly increasing computation. Performance (AUC score in this case): There is typically an optimal depth that balances underfitting and overfitting. A scatter plot showing the correlation between tree depth and the AUC metric will allow the ML Specialist to: Identify whether increasing depth leads to diminishing returns. Choose an optimal tree depth that balances performance with training efficiency. Reduce the search space of hyperparameters, speeding up tuning and lowering costs.

A. No, doesn't help to set/reduce hyperparameter value/range B. No, honestly this is gibberish to me C. No, doesn't help to reduce hyperparameter value range D. YES, this help me understand how to set max tree depth hyperparameter

Option C. See it is doing a scatter plot on the metric for each iteration. Each iteration is running with a certain set of hyper parameters. So if I plot this. and I find which iteration has the best metric, I could simply pick up those set of hyperparameters. D will only led to the tuning of maximum tree depth. I am not sure which option would satisfy the goal to decrease cost but just looking at maximum tree depth doesnt seem right to me. It might be a way to just look at the tree depth and tune just that parameter and since you are only tuning 1 paramter, it may be cheaper, but would that lead to a usable model? I think it should be option C.

On what basis the correct answers are provided in this platform? Are they assuming this is the correct answer or it is taken from somewhere ?

D IS THE CORRECT

Option D, can also be useful in hyperparameter tuning for tree-based ensemble models, especially if the maximum tree depth is one of the hyperparameters you want to optimize. However, when the goal is to decrease training time and costs by reconfiguring input hyperparameter ranges, a scatter plot showing the performance of the objective metric over each training iteration (Option C) is generally more directly related to the hyperparameter tuning process. It helps you track how the model's performance changes during hyperparameter tuning, which is critical for making decisions about which hyperparameter ranges to explore further. Option D is valuable for understanding the relationship between maximum tree depth and the objective metric, but it might not provide as comprehensive insights into the overall hyperparameter tuning process compared to Option C.

A. NO - it is about data discovery B. NO - it is about data discovery C. MIGHT - (NO) is a training iteration the overnight training the question is referring to ? (YES) Or each HPO training within each night ? D. YES - the less ambiguous answer

I think that C should be the right answer. The specialist can monitor how the model works by changing hyperparameters' values in each training iteration.

Option D

A and B are wrong, because is totally out of question context. C is for monitoring a model, it doesn't help to change your HP range. D is the only answer that applies to the question.

I think it should be c

By plotting the performance of the objective metric (AUC) over each training iteration, the Specialist can analyze how different hyperparameter configurations affect the model's performance. This visualization helps in understanding which hyperparameter combinations lead to better results and allows the Specialist to identify areas of improvement.

D: By analyzing this relationship, the Specialist can adjust the range of maximum tree depth values used during hyperparameter tuning to decrease training time and costs.

D Seems like the best answer. When answer is considered correct who is making that call an is there any justification provided for us to learn from?

It's about parameters, not about dimensionality.