Today we are faced with a wide range of data analysis tools.
These tools vary from solutions that try to automate an action (anomaly detection, classification, prediction, optimization…) to solutions that allow humans to interpret complex and/or massive data.
Each problem has its own particularities and these determine that some tools are more suitable than others for its treatment.
Understanding the needs of a given application in a particular domain is a real difficulty that requires experience in Data Science. Additionally, each domain has its own characteristics regarding the data is (and is not) available, the frequency with which it can be gathered, and the complexity and cost of this collection.
Setting up a practical data pipeline (chain of processing steps) requires a deep understanding of the problem at hand (domain knowledge).
At Codas Lab we create new data analysis tools in domains in which we have years of experience and in collaboration with experts from each domain.
Our tools can be machine learning solutions —which mainly use multivariate techniques—, visualization tools for data exploration, and statistical inference tools where significance can be tested.
In order to optimize the use of these tools, we organize practical courses in which we teach how to avoid subtle pitfalls —so common in Data Science— and how to maximize the information collected.
And in the same way, we disseminate and share these tools with the community (you can learn more about them by clicking here).
For a data solution to be successful it must have both the knowledge of the domain expert and the Data Science expert, and this is a fact that is sometimes forgotten in the scientific community.
It is common for domain experts to use public software tools without a good understanding of their practical limitations. And likewise, data experts often apply their tools without a proper understanding of the domain.
At Codas Lab we firmly believe in the importance of both facets and the need to know how to combine them.
If you want to receive more information about our Development of new machine learning and statistical tools reseach line, do not hesitate to contact us.
