728 x 90

Machine Learning: ML

Machine Learning: ML

Machine learning has become an essential component in recent years of business productivity creating a market alone of more than 8 billion euros worldwide. Data, the gold of the 21st century, is the engine of this new technology. It brings a lot of value in industries such as health or education but also in environments

Machine learning has become an essential component in recent years of business productivity creating a market alone of more than 8 billion euros worldwide. Data, the gold of the 21st century, is the engine of this new technology. It brings a lot of value in industries such as health or education but also in environments where it is less obvious to see its impact horizontally in the consulting profession. You only need to take a look at the data job to understand that the job market is undergoing a transformation and that consultants must also master this technology.

Data and machine learning… these are two buzzwords that we use indiscriminately… when in reality they are two sides of the same coin, not only complementary but inherent to each other. This is why it is important to define machine learning, the actors in its value chain, and how data feeds this process. Today, an automation project brings together certain technologies that allow a machine to learn from past examples to explain an event (descriptive goal) or anticipate a future event (predictive goal).

To define machine learning, we use this analogy with strategic approaches: Top down vs Bottom up.

Historically, an engineer would use the Top-down approach in order to achieve a sufficiently sophisticated result of automation by establishing a system of rules that the machine recorded and followed to the letter. It started with pre-established rules and gradually moved down to a predictive model.

Data + Logic = Output

Today, machine learning is completely rethinking the way we think about automation and learning. This is the bottom up approach. The machine starts with the results, the data, and the experience to deduce by going up a model that will make it possible to carry out these predictions.

Data + Output = Logic

In order to popularize these concepts and to approach the data-machine learning relationship in a pedagogical way, we have therefore decided to break down the actors of this ecosystem here in an article: the “ parameters and technologies ” which correspond to the raw material, the “ Orchestrators ” who are the tools and finally the products. This technique will allow us to explain their roles within the chain.
Regardless of the infrastructure used (traditional pipeline vs turnkey platform), it’s the same process:

  1. Data is collected via acquisition tools if it is external to the company or by connecting the different applications (CRM, ERP) via APIs (this is the means of communication between two programs) to establish a lake or a database.
  2. We clean, prepare and structure this data to make it usable by an analysis model: the models are often provided in open-source (especially in turnkey platforms) but the difference and the intelligence of effective learning lie above all in the creation of relevant “ features ” (these are the parameters used by the model that the engineer will choose) via the upstream preparation of this data. This step is important and it is often a very big point of differentiation.
  3. Once the data has been prepared, the model is chosen according to the desired use case, and the model is trained on this data: this is the learning phase.
  4. Once the model is trained – we deploy it as a product
  5. The user helps to improve the relevance of the product because it provides data that crosses the chain and allows the model to be trained continuously via the network.

The more the user uses the product, the more qualified data he provides, the more the algorithms become relevant, the more the product improves, the more the user uses the product… Machine learning has become the cornerstone of a project of automation and support from many service companies. But before using it, you still have to understand where the value hides, how and where to use it.

Related Posts

Leave a Comment

Your email address will not be published. Required fields are marked with *