Uncategorized

python – Perhaps it is due to parallel computing that machine learning produces different results after taking the same random seed. How can this be resolved?


Using sklearn.model_selection.RandomizedSearchCV function. The parameter list was taken after using np. random. seed (42) and should be fixed. And every step has added random_state=42. ShuffleSplit used for function partitioning;

Some code snippets are as follows:

**my_seed = 42
RandomizedSearchCV(estimator=model, param_distributions=param_distributions, n_iter=Random_search_n_iter_list[i],
cv=ShuffleSplit(test_size=0.3, train_size=0.7, n_splits=10,random_state=my_seed),random_state=my_seed,scoring=”accuracy”) **

So I guess it’s because parallel computing causes machine learning to have different results after taking the same random seed. How can we solve this problem? Or did I make a mistake in the random process itself? Could you please inform me, big shots

The parameter list was taken after using np. random. seed (42) and should be fixed. And every step has added random_state=42. ShuffleSplit used for function partitioning;

After researching the information, it was found that parallel computing may lead to different results in machine learning after taking the same random seed



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *