Is there a possibility for numpy to replace a for-loop for performance improvements?

Question

Is there a way to improve the performance of this code with numpy or python in general? The goal is to build a trainings set. features is the raw data. I want to us