Is it logic (possible) to get %100 in accuracy and classification rate?

I have trained my model with LDA (I have 3 different train datasets ).for each one classification rate and accuracy =%100

First Training 30% Training %70 Testing Second Training 70% Training %30 Testing Third Training 30% Training(Random) %70 Testing

[John D'Errico]

No. No method can do this. At least not unless your data is itself perfect, with absolutely no noise.

[Bajdar Nour]

thanks Dear @John D'Errico, I use Fit discriminant classifier. If you know please tell me where is my mistake?