What spiega is about
- Knowledge sharing
- Portfolio
data modeling
In modern data platforms a proper data models decides the overall efficiency and cost of the infrastructure.
A proper setup on how the information is stored allows to have efficient data pipelines, fast batch processes, clever retention policy, faster insight creation and cost reduction.
Data models are designed for interoperability between services and require clear focus on business cases to act efficiently.
A data model design requires the collection of requirements of many business units and understand their operations.
Data models need to be designed to allow expansion without suffering from backward incompatibility.
Design a proper data model requires a lot of skills in one:
- collect and understand requirements -> product management
- understand the tech stack and it's components -> devops
- understand business units and their needs -> product management
- understand the data and the transformation cost -> data analytics
- understand the insights generated by data and set up KPIs -> data analytics
- feature engineering and preprocessing -> machine learning
- propose a correctly sized infrastructure -> solution architect
Data modeling is crucial for a proper product design, operation efficiency, analitical consistency, error search. A good example of data modeling is (fhir)[https://www.hl7.org/fhir/index.html] which is widely used to share patient's health information.
taxonomy
Taxonomy is an important process to form a data model. The company thinks about its own structure, the customer personas, audience segments, product portfolio...
relational
For historical reasons the most used databases are relational which makes
nested
protocols
fhir
Fhir is a widely used protocol to store patient's information.