Data de publicação: 03/05/2023
Quite recently, considerable attention has been paid to developing
artificial intelligence and data science areas. This has been driven
by scientific advances and the growing number of software and
services that are popularizing machine learning techniques and
algorithms and driving people with less knowledge in areas such
as statistics and mathematics to create their predictive models. As
a result, the machine learning field is no longer only scientific
and has aroused the interest of companies from different domains.
These events led to the emergence of multiple tools such as Scikit-
Learn, Tensorflow, Keras, Pycaret, and a vast number of cloud-based
machine learning services that provide an acceleration in the development
of predictive models at speeds never seen. However, many
challenges remain in operationalizing and maintaining machine
learning-centered products, making many business initiatives frustrated.
In this scenario, practical experience shows that machine
learning is only a slice of a more extensive set of practices and
technologies necessary to build solutions in this area. In this paper,
the main goal is to identify the challenges currently faced by data
scientists in developing Machine Learning-centric products and
how Machine Learning Operations can support overcoming them.
For this purpose, a survey was conducted that collected answers
from 66 Brazilian professionals in data science. From the challenges
identified, the importance of Machine Learning Operations practices
as an integrated part of the Machine Learning lifecycle was
explored. Finally, this work contributes to filling the gap in Machine
Learning Operations in daily activities involving data science and
advancing this research field in Brazil.