Improving Delivery Precision with Machine Learning : Identifying key factors affecting delivery precision and utilizing them in machine learning models to identify orders at risk

The COVID-19 pandemic have clearly highlighted the value in a pharmaceu-tical industry that can respond quickly to the market and deliver with highprecision to support ambitious action plans against biological threats. Thisthesis attempts to identify key factors that affects the delivery precisionfor Cytiva which is a global leader within the life science industry. Thesefactors include everything from order information and stocking policies toshipping regions and seasonality variables. Additionally, this thesis exploresthe possibility of using machine learning models to identify orders at risk ofbecoming late at en early stage and thereby allow for preventative actionsto be taken in time. The statistical analysis and modelling are performedusing historical order data generated during the years 2018 to 2020.This thesis concludes that the Random forest model is the best per-forming model for this use case. Although the current data set did not allowfor any of the models to reach quite the performance scores that wouldrender them useful in the everyday operations of the company, severalimportant factors were identified. These critical factors can be used toidentify negative patterns in the factory and to streamline the production