Data Mining application case study: Pharmaceutical company.
Today we would like to share with you a success story we delivered for a leading Logistics Operator in the pharmaceutical sector.
To set the scene, the main characteristics were as follows:
– Certified under GDP/GMP regulations
– 20,000 m2 and 4 floors (height 30 m)
– Investment of 33 million Swiss francs
– Supply from factory or suppliers’ warehouse
– +6,000 end customers
– Automated warehouse: 6,000 pallets and 66,000 boxes
– Inbound rate: 100 pallets/day
– Outbound rate: 2,500 order lines/day
– Internal transfers: 1,000 order lines/day
In this project developed by LISDATA, one of the main objectives was to analyse the cold chain: there were antibiotic vaccines within a range of -20°C to +8°C, and anything outside that range reduced the quality of the medicine or vaccine. In addition, there were containers available that guaranteed the temperature range for 24 hours.
Challenge: “The reliability of the cold chain is unknown”.
It is necessary to identify the origin of the breaks and the potential impact they generate for customers, in order to be able to claim the product as quickly as possible; in other words, to ensure traceability.
Data Mining solution:
Given the complexity of the analysis and the number of variables involved (+100) in the issue, it was decided to carry out a study using data mining techniques to reduce the number of variables involved.
A cold-chain loss was defined as any movement that, after leaving the cold warehouse, exceeds 24 hours before dispatch.
We can see how, thanks to simple data mining techniques, we can reduce the problem by focusing on variables such as priority, day of the week, or operator.
Data Mining process.
A node-based system was developed that made it possible to automate the merging and cleansing of more than 180 Excel files.
Thanks to different data mining algorithms, the variables involved in the cold-chain break issue were identified, and through “machine learning”, an intelligence process was created that, by learning from historical data, identifies potential breakpoints when new data is received.
Poor management of order priorities and extensive bureaucracy carried out outside the cold warehouse were determining factors in the cold-chain failure!
“Thanks to the combination of data mining algorithms and visualisation tools (Business Intelligence), we were able to keep a critical process under control, avoiding errors that could be fatal for our company.”