How could we further improve the threat scoring algorithm? Can we use the previously discussed threat scoring code with a Python-based scheduler?