Hi,

The editor's kernel remembers the variables defined by all scripts run in it. Resetting the kernel clears that, and so the new results are based only on your actual script.

Can you share more information or the code, so that I can take a better look at the problem?

Regards,

Patrick

Hi,

The editor's kernel remembers the variables defined by all scripts run in it. Resetting the kernel clears that, and so the new results are based only on your actual script.

Can you share more information or the code, so that I can take a better look at the problem?

Regards,

Patrick

Patrick Timmons thank you for answering , of course please find bellow the code

You may need to use the seed() method. Random seeds are also common to create reproducible results for research. For example, if you're a data scientist and you want to publish your results with some kind of model that uses randomness (e.g. a random forest), you'll want to include a seed in your published code so people can make sure your calculations are reproducible

Mayibongwe Dube as i understand from your answer sir, my results are varying depending on the choice of training and testing data , and I should use the seed function to fix the same train test data every time , and yes I want to publish my results.

Mayibongwe Dube is right, you need to set random.seed in your code, and set random_state for the classifiers which support it to get reproducible results.

For the dataset, another idea is to set a random seed, but then save and pickle the output of the dataset train/val/test splits, and use the saved splits in future. This also makes it easier to later publish the dataset of which samples belong to which splits.

Myriam El Mesbahi

Best.