Every DATA SCIENCE project is a challenge in itself.
The possibility of being able to translate reality into a magic formula that can emulate and reproduce it is tied to the complexity of the different quantitative and statistical methods behind the MACHINE LEARNING algorithms that are required.
This result is also tied to the Analytical Approach of the Data Scientist, to the possibility of having a Data Set that has the correct volume and representativeness or to the possibility of collecting them, in order to later be able to develop a successful PREDICTIVE MODEL.
However, when a significantly large data set is analyzed, it is very easy to find variables or phenomena that are related to the naked eye, although in reality it is only a fortuitous result. A perfect model may actually be hiding arbitrary correlations that can only be identified by those who know in depth what happens in the REAL WORLD.
For all this, the biggest challenge in DATA SCIENCE is being able to work together with the CLIENT. Nothing replaces INSIGHT and the context information that someone who has been in the same industry for 15 years, in the same activity or engaged in the same process can have.
There is a popular phrase in DS that indicates that the correlation of variables does not imply causation and on the SPURIOUS CORRELATIONS website we can find thousands of examples on this subject.
At FERSYS we work together with our clients carrying out a CONSULTATIVE DISCOVERY jointly and involving key users in EARLY STAGES of the project in order to obtain the best results.
