High phosphorus levels in steel result in products with poor mechanical properties, erode profit margins, lower flexibility in downstream processing and disrupt delivery schedules. A robust advanced analytics model was developed for finding optimum process parameters resulting in low end blow phosphorus and high phosphorus partitioning in an LD shop. Best patterns were determined for oxygen blowing, material charging sequence and bottom stirring. This model, along with a machine learning approach, can suggest real-time operating recipes to the operators resulting in a massive 13% increase in the current success rate of the heats complying with the aim phosphorus level.