I have the following problem in Python : I have a dataset with 7 features and 4 classes. I used a MLP classifier and I got very good results in terms of accuracy. Now, they are asking me to demonstrate if it was really necessary by using a MLP instead of a simpler classifier such as a linear one. So, I would like to plot some graph taking two by two the features and show that they can't be linearly separated by linear regions. Is there any library that can do that without writing a lot of code's rows ? More general, would it be possible to plot all 7 features(instead of pair of features)on the same graph and show that is not possible to face this problem through a linear classifier?
Asked
Active
Viewed 204 times
0
-
SVMs can learn nonlinear decision boundaries (using kernels). – cheersmate Nov 06 '18 at 08:19
-
Ok, I'm pretty new on that.Could you guide me to answer my question? Thanks – Alex Nov 06 '18 at 08:45