search

ai-se / surrogate

surrogate modeling. moea. decision space
MIT License
0 stars 0 forks source link

check SMOTE #16

Open vivekaxl opened 8 years ago

vivekaxl commented 8 years ago

@timm @Ginfung : Suggestions

I have tried two strategies for dependent techniques:

        # For dependent Variables
        d1 = euclidean_distance(new_indep, case.independent)
        d2 = euclidean_distance(new_indep, x.independent)
        new_dep = (d2 * case.dependent + d1 * x.dependent) / (d1 + d2)

2.

        from random import random
        new_dep = case.dependent + random() * (x.dependent - case.dependent)

The results are very similar (both are equally bad)