2022’s top digitization trends: what will define the analytics space this year?

Pedro Vale, Chief Technology Officer, LYTT, January 21 2022

When you’ve worked in computing for over two decades, you gain an appreciation of how dynamic the sector is. There has never been a more exciting time to work with data, and since joining LYTT in September 2021, I have been involved in many projects that have the potential to transform operations across multiple industries, such as Carbon Capture Storage Monitoring, Power & Utilities, and Oil & Gas.

At the beginning of my career in the early 00s, when I worked on business process management platforms and mobile user interfaces, there were no smartphones. The volume of data we routinely work with now was unthinkable back then. I’ve been privileged to work at the forefront of industrial data analytics during a period of rapid innovation, at both start-ups and large corporations.

After managing a 60-person team and working in the Big Data space, increasingly touching on the related fields of the Internet of Things (IoT) and sensor analytics, I was ready for a new challenge. LYTT provides that challenge as it delivers real-time, actionable insights to operators using its cutting-edge, artificial intelligence driven data analytics platform.

With demand for new ways to optimize operations growing rapidly, we have doubled the size of the software and data team at LYTT since I joined. This year, we are set to double it again as we push the boundaries of LYTT’s platform and help operators accelerate their digital journeys and make the most of the opportunities available in the energy transition.

My top three trends in digitization for 2022

We are seeing a new wave of applications for machine learning, driven by an ever-increasing variety and quantity of data gathered by all sectors. Previously inconceivable volumes of data are used to train algorithms and models, leading to more accurate insights that enable smarter, more agile operational decision-making.

At LYTT, we are always innovating to ensure that our own software and data analytics solutions remain cutting-edge. The three areas below have become more relevant as LYTT scales in order to meet our customers’ evolving needs, and in 2022, they will continue to grow in importance.

1. Machine Learning operations (ML Ops)

Machine Learning algorithms don't stop evolving when they are used in real-world scenarios. They undergo a continuous process of improvement and refinement as new data becomes available, as computing power increases, and as the requirements of specific use cases evolve.

ML Ops is about managing machine learning models efficiently during their lifecycle with regard to individual applications of the algorithms. This includes bringing the models to production – but also using and maintaining them.

Machine learning models will ideally have lifecycles completely independent from the applications that use them. We are already implementing this at LYTT, as it gives Data Scientists greater flexibility to update algorithms and ensure that ML models continue to positively transform operational decision-making for end users.

2. Digital twins

A ‘digital twin’ is not a new concept. Stemming from the IoT space, this tool enables operators to digitally represent a physical asset, providing a virtual model of the asset for simulations. The model aims to replicate the real world and provide a testing environment for operational decisions.

In industries such as Oil & Gas, digital twins are often used for predictive maintenance – to model likely scenarios and predict the outcome of possible remedial actions. However, using digital twins can also unlock a more accurate understanding of why specific types of data should be acquired, allowing us to inform machine learning algorithms using the most relevant data and, in turn, make better predictions.

Physics calculations can also be used to derive extra sensors based on real sensors, by extrapolating from existing datasets to predict what additional data-streams might show. This allows for extra data points to feed into the algorithms, boosting their accuracy further. This ultimately leads to optimized operational decision making and reduced risk.

3. Data mesh

Data management has evolved significantly in recent years – and now looks set to turn a new corner with the advent of ‘data mesh’ architecture.

The classic business intelligence approach to organizing and using data was the ‘data warehouse’. This enabled large organizations to integrate key datasets and extract value, but often led to inconsistencies across departments.

In recent years, this has evolved into ‘data lakes’, where all data is dropped into a single large database, with specialized tools used to make sense of that data and access insights. Data lakes, however, can be impractical for non-specialist users due to the volume of data, making it difficult to extract specific information.

As a result, we are increasingly seeing greater adoption of ‘data mesh’ architectures. This methodology takes a decentralized approach with distinct domains of data. The focus is not on ingesting data from several sources, but instead is about making data available to other applications in a way that is domain specific, and therefore more useful.

As LYTT integrates into broader software ecosystems, with customers that may already have data management platforms in place, our ability to add our end-to-end solution into existing application infrastructure to deliver scalable insights ensures a seamless process for customers.

What next for LYTT?

Driven by a talented and dedicated team, LYTT has gained significant experience solving complex operational challenges in Oil & Gas, empowering customers with proven solutions. It became clear that the underlying algorithms are readily transferable to multiple applications to many sectors.

When a particular use case is considered, it is still firstly a software and data engineering problem. The models that LYTT developed for Oil & Gas to solve challenges related to flow, integrity, solids and seismic can be translated to other subsurface industries. For example, our model is capable of detecting leaks in water and sewage pipes when trained with relevant data.

No matter the industry, or the challenge, instantaneous and previously inaccessible insights will empower operators to make faster and smarter decisions.

Join LYTT’s journey, and my team, and help transform data into actionable insights! Our latest vacancies are available here.

This website uses cookies. Click here to learn more. That's Fine