Problem:

Lack of information about predicted parcel load for distribution network hubs. Information about the number of parcels currently moving between network hubs is available in the system, but it is hard to retrieve for business users. Lack of data visualizations, that are clear for business users. The data volume of several TB, reporting data should be delivered to users within one hour.

Solution:

End-to-end solution has been implemented in the scope of the project. It covers the whole data processing flow: starting with source systems integration, through data transformation to semantic model and predictive modeling, up to results visualization for end users. The solution has been implemented with Microsoft Azure services (PaaS model) and Power BI. The main purpose of PoC implementation was to prove the applicability of predictive modeling in the area of parcel load forecasts reporting.

Business Impact:

The goal of the solution is to improve the work efficiency of distribution network hubs. It can be achieved by delivering clear data visualizations with accurate information about predicted parcel load for each hub in the dimension of time. Reports should improve planning processes for human resources and the workload of sorting machines.