Systems and methods for predicting that a stock keeping unit (SKU) is likely to be scrapped and performing preventative actions accordingly. The system receives a request to initiate a process related to predicting scrap likelihood associated with a plurality of SKUs, inputs data associated with each SKU into a trained prediction model, outputs, from the trained prediction model, a prediction value for each SKU, measures the trained prediction model by calculating a set of values, and generates a matrix comprising the calculated set of values. Using the matrix, a threshold prediction value is determined, and each SKU prediction value is compared to the threshold prediction value. Based on the comparison, a SKU of the plurality of SKUs is predicted as likely to be scrapped and at least one of a plurality of preventative actions is performed.