Kernel methods may take several order magnitude more time than regressors.
I propose to introduce a time threshold (t) seconds for training algorithms. When the training time of an algorithm surpasses this threshold, the algorithm's training is terminated, and lazypredict goes to the next algorithm.
We may imagine those following implementation details:
All training algorithms are encapsulated in threads to facilitate time management.
API exposes to the user a time threshold parameter (t).
If the training time exceeds 't', the corresponding thread is terminated, and lazypredict goes to the next training loop.
Kernel methods may take several order magnitude more time than regressors.
I propose to introduce a time threshold (t) seconds for training algorithms. When the training time of an algorithm surpasses this threshold, the algorithm's training is terminated, and lazypredict goes to the next algorithm.
We may imagine those following implementation details: