0
$\begingroup$

This is what I have done :: divided my dataset into training and testing sets--> got significant features via. feature selection using sequential feature selector ( MLxtend) on the training set--> used significant features to fit a linear predictive model--> test the model on the testing set.

However, the error for both my test and training set is not small. Should I reconsider the decision to split my dataset? Any help is much appreciated!!

(Not a data science person)

Improve this question
$\endgroup$

1 Answer 1

Reset to default
0
$\begingroup$

It is not clear the ratio that you have used to do the train-test split. If the dataset is small, you can try with cross-validation and check the metrics. Also, can try with various rations 80-20, 30-70 and compare the metrics. Post this, you may also check using different techniques for feature selection - Lasso, tree-based etc.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.