Business intelligence on dairy production data for DeLaval Argentina

DeLaval Argentina supplies technology to the dairy sector. Milking and animal health systems permanently record each animal: how much it produces, when it calves, what health events it had, when it is culled.

Having data and being able to decide with it are different things: that record lives in operational tables built for the day to day, not for answering business questions.

• Case by case analysis: you can look at one animal, not at the herd's evolution over time.
• Isolated domains: production, health and reproductive events cannot be crossed without leaving the source system.
• Structure tied to operations: any analysis depends on how the operational system chose to store things.

The cost is that the information exists and decisions keep being made on experience, because nobody can ask the data what they need.

We consolidated that information into a dedicated dimensional model and took it to the cloud, so the herd can be analyzed as a whole rather than case by case.

It is the analytical complement to the infrastructure and support work the company already had in place.

A model built for asking

• Dimensional schema consolidating the livestock business domains into analytical tables split by axis.
• Active animals, culls, health events, abortion events, calvings, milking records, animal groupings and clients.
• Each dimension can be crossed with the others without depending on the source system structure.

Extraction and normalization

• Data taken from the dairy management systems and from diagnostic files.
• Normalization of the different origins into a common load format.

Destination in the cloud

• Export of the dimensional set to the company's cloud platform.
• The model is left available to analysis and reporting tools.

Architecture and technology

• Dimensional model over the livestock business domains, with normalized loading from multiple origins.
• Publication of the set on the company's cloud platform.
• Analytical tables split by axis, crossable without passing through the source system.

Learn more about software development.

The daily record of each animal became a series that can be analyzed by group and by client.

Starting point

• Production and health data in the milking system operational tables.
• Analysis possible only case by case.
• Business domains with no way to be crossed.

Strategic impact

• Milk production, herd health evolution, reproductive events and culls are tracked together.
• Analysis works per animal and aggregated by group and by client.
• The model does not depend on the operational system structure, so it survives its changes.
• It is available in the cloud for whichever analysis tools the company chooses.