< back

Top 10 best practices for machine learning, NLP and IA

There is a lot of excitement  about artificial intelligence (AI), machine learning (ML), and natural language processing (NLP). Although many of these technologies have been
available for decades, new advancements in compute power along with new algorithmic developments are making these technologies more attractive to early adopter companies.

  1. Know your business problem. It is important to start with a real business problem and clear objectives when embarking on a more advanced analytics project.
  2.  Start somewhere. It is important to start somewhere. Change is happening fast and organizations cannot afford to be complacent.
  3. Consider the pros and cons of open source. Open source is rapidly becoming a go-to software for machine learning, NLP, and AI. Some of this has to do with the large community around open source as well as the fact that it is an inexpensive way to get started.
  4. Hire some data scientists. There are excellent tools on the market that can help everyone become more productive. However, the reality is that if your organization really wants to do sophisticated analytics, it is probably going to have to hire at least a few data scientists.
  5. Build a center of excellence. As described above, a CoE can be a great way to ensure the infrastructure and analytics you implement are coherent. CoEs can help your organization
    disseminate information, provide training, or maintain governance.
  6. Hire an analytics guru to be in charge. Often the best person to be responsible for an analytics effort is someone who really understands analytics. Their titles may vary—some organizationshire CAOs and others hire VPs of analytics.
  7. Think about the architecture, including the cloud. The data warehouse is not going anywhere anytime soon. However, machine learning, NLP, and AI may necessitate moving beyond the data warehouse to platforms that can support multistructured data and iterative analytics.
  8. Pay attention to data quality. The issue of data quality is also front and center for many organizations. The notion put forth by some “experts” that quality issues would “wash out” with
    big data is not true.
  9. Operationalize your analytics. Analytics provides the greatest amount of value when someone can take action on the results. That action can take many forms.
  10. Think about the success cycle. We have seen that success begets success. If you are successful
    with one project, do not stop there.

Source : TDWI Advanced Analytics: Moving Toward AI,Machine Learning, and Natural Language Processing Report

Credit photo : www.pexels.com


Vous êtes membre?

Utiliser la section Espace membre pour modifier votre profil d’entreprise ou consulter les projets d’automatisation affichés.