0
$\begingroup$

I am running statistical test to measure the difference between two fields (graduation age women and graduation age men). I am using ttest_ind() and I got the following results:

Ttest_indResult(statistic=10.965703359355636, pvalue=5.7138670616970964e-16)

From what I read (I have not a strong statistical background), the p-value is significant. In terms of results for features selection in machine learning algorithms, what would it mean? Currently I am considering graduation age as a feature that can actually improve my ML (model performance metrics have been improved after adding it).

Improve this question
$\endgroup$

1 Answer 1

Reset to default
1
$\begingroup$

the t-test tells you that the average graduation age of men is statistically significantly different than the average graduation age of women, under all the usual assumptions of the t-test. For feature selection, it tells you that if you're trying to predict gender then a graduation age will be informative or vice versa. Although it does not tell you anything about if the feature will improve your model if other features are already there. It does not tell you anything about improving a model performance for predicting anything else but gender or graduation age

Improve this answer
$\endgroup$
2
  • $\begingroup$ Thank you so much rep_ho. So the result that I got would suggest that there is significantly difference between the two ages, am I right? $\endgroup$
    – Math
    Commented May 10, 2021 at 14:57
  • 1
    $\begingroup$ between two sexes in the graduation age $\endgroup$
    – rep_ho
    Commented May 10, 2021 at 15:02

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.