-2

I am working on a transactional dataset where I am trying to create a churn prediction model using SVM. I have manually defined my training and testing dataset and have not used CrossValidation.

However, to find the best hyperparameters for my SVM(svc) model, is there any alternative way to do it without Grid Search CV, my objective here is to try and prevent any data leakage happening as I understand that using CV wouldn't solve that problem.

Kiedi7
  • 51
  • 1
  • 8
  • 1
    Using `"using CV wouldn't solve that problem"`? Please explain. GridSearchCV use ParameterGrid to expand the supplied params to all possible combinations, but no data leakage happens there, if used properly. Please explain more about your scenario. – Vivek Kumar Apr 23 '18 at 05:44

1 Answers1

2

Parameter grid may help you to populate all the different combination of parameters ParameterGrid .

Laurent R
  • 783
  • 1
  • 6
  • 25
  • Thanks for this, I was able to use ParamaterGrid and create a loop to iterate over! Appreciate it :) – Kiedi7 Apr 22 '18 at 20:17