In big data, skipping the necessary steps is counter-productive

What is the key to making data science, business intelligence or data intelligence initiatives successful?

A holographic interface.

An increasing number of fashion companies want to take advantage of the phenomenal potential of data.

Those who launch into the subject often tackle it through data sciences. Essentially, they approach the topic in reverse. At Lectra, we think that it is best to lay solid foundations before any data competitive analysis of fashion industry even occurs.

Frequently, a company that launches its first projects around data is looking to make its financial data visible in a new way. It is often the financial team (with the involvement of the CFO) that adopts data science in order to rapidly share the benefit of projects and see positive results.

If an initial data project is successful, a company may then be tempted to employ data science in similar projects with other teams, using software, systems and machines to visualize information differently and share it with as many people as possible.

It is often at this stage that the first issues appear. If a company has not put in place rigorous data governance, the types of data arriving in the systems can vary with each update, making them unusable. Garbage in, garbage out.

What is the key to making data science, business intelligence or data intelligence initiatives successful?

Data quality is an absolute prerequisite

At Lectra, we see it as essential to ensure the consistency – and therefore the quality – of the data pipeline, before any analysis takes place. It is a challenging task: 90% of web data is unstructured – making it necessary to commit nearly a third of a data scientist’s time to continuously cleaning, qualifying and organizing data so that it can be reliably exploited.

Without quality data, how can we rely on machine-learning algorithms to enable decision-support tools to deliver accurate analyses and relevant proposals? The machine's intelligence is completely undermined if it does not understand the data it receives.

Data engineering and the use of data quality systems are essential. We believe that a scalable and secure data architecture will be the foundation of data science tools and projects of the future. At Lectra, we see this as a prerequisite for the implementation of any effective artificial intelligence – we strive to control the quality of the data as soon as we receive it.