- Permutation importance: https://www.mathworks.com/help/stats/permutationimportance.html
- Shapley values: https://www.mathworks.com/help/stats/shapley.html
Determining Coefficient Weights in a Rational Quadratic GPR model
1 次查看(过去 30 天)
显示 更早的评论
Hello there I have a trianed GPR model stored in the workspace as "GPRModel". I have 63 input featured for my trianed model and I am trying to detmine which features are weighted the highest and contribute the most to the model's predictions. My training data with the features is stored in a Table named: "AllFinalTrials". Let me know if there is any way to determine the coefficents or wieghts of the featured for this Rational Quadratic GPR model. Thank you so much!
0 个评论
回答(1 个)
Drew
2024-4-3
If you want to know "which features ... contribute the most to the model's predictions", one way to do that is to use model interpretability techniques. Here are a couple of interpretability techniques that can produce an importance value for each feature of your Rational Quadratic GPR model:
For a general introduction to model interpretability, see https://www.mathworks.com/discovery/interpretability.html
If this answer helps you, please remember to accept the answer.
5 个评论
Drew
2024-4-3
If you have a MATLAB license, you can get access to R2024a in two ways:
(1) Access R2024a online at matlab.mathworks.com
(2) R2024a is also available for download from matlab.mathworks.com
Drew
2024-4-12
Since you wrote "Thank you", I recommend to accept the answer. I see you have asked 9 questions, but never accepted an answer. If you are wondering how to accept an answer, there should be an "Accept this answer" button that appears next to the answerer's name, near the top of the answer.
另请参阅
类别
在 Help Center 和 File Exchange 中查找有关 Portfolio Optimization and Asset Allocation 的更多信息
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!