This repository contains different data science solutions that can be used in different data science projects with different applications.
This solution contains python code to calculate the rolling cosine similarity over time series data. The solution is implemented using Numpy and Numba. The notebook contains detailed explanation to the approach as well as the implementation using Numpy and Numba. This solution isn't new, there exist different approaches, however, I have added the rolling window capability as well as modifying the numpy code so that I'm able to use numba for faster computations. I have evaluated different solutions and showed that using numpy with Numba is the fastest among all. The link for the notebook can be found here Rolling Cosine Similarity