Data Analytics

Introduction: 

Data Analytics has traditionally been associated with the processes involved in using data to inform decision making. It builds on the underpinning principles of data management that are required to build any kind of IT system, including the integration of IoT operational and back-end business systems.

In the context of IoT, Data analytics encompasses many approaches including big data, in-memory computing, cloud computing, NoSQL databases, data integration, and interactive analytics, as shown in the diagram below.

Diagram courtesy of Jorge Lizama. GHD

Historically, data analytics took the form of Decision / Executive Support Systems starting in the 1970s, then evolving into Online Analytical Processing (OLAP), Business Intelligence (BI) in the 1990s.  

It is common to think of data analytics in terms of the volume, velocity, and variety of the data. Volume refers to the quantity of data, velocity to the speed at which the data is generated, and variety to the different types of data. Over the past few years, two new Vs, value and veracity have been introduced. Veracity refers to the quality of the data, and value refers to the benefit that the organisations can gain from the volume and variety of data that is being delivered with great velocity, if they are able to depend on its veracity.

Diagram courtesy of Arthur Baoustanos, aib Consulting Services

The current approach to managing data collected from IoT devices is to sense/observe the data, move it into the cloud, process and analyse it there, visualise it for decision making purposes (using technologies including augmented reality), then either store or discard it partially/completely.

In recent times the exponential growth of data has created situations where "traditional" analytical methods are not viable and the term big data analytics is being used to describe new analytical techniques developed to cope with these situations. Big data analytics is often associated with IoT because many IoT applications involve large numbers of sensors generating large volumes of data. Also, many IoT applications involve the integration of a large variety of data formats such as weather data, machine vision and the like. 

A key challenge of IoT systems that generate or integrate a lot of data is how to make sense of it and how best to make use of it. This is driving the uptake of cognitive computing systems that assist analysts in determining insights and drive outcomes not possible with traditional analysis.

Planning for data analytics

The critical questions that organisations will need to answer when embarking on the journey to advanced data analytics are:

• Where does the organisation want to go (goals)?
• How will we get there?
• What do we need to get there?
• Will our current structure allows us to get there?
• What changes do I need to make to get us there?

It is important to start with the business objective: define critical business issues and decide where value will be derived. Then evaluate which data is required to assess the identified issues and determine any gaps in relevant data. Be as specific as possible about what decisions the company will make based on that information. Departments and divisions within the organisation should collaborate to understand exactly what information is required to address common business goals. Data could also be purchased from outside sources to complement internal data collection.

Once an organisation has decided to optimise their efficiency using data analytics, they should look at long as well as short term goals, and set specific efficiency or process change targets in order to get the most out of their investment as shown in the diagram below.

Diagram courtesy of Umesh Bhutoria, EnergyTech Ventures

A gap analysis of people skills (users, engineers, managers), data (points and frequency) and investment needed to reach goals should also be carried out, to ensure that all stakeholders are willing to see the process out through trials to implementation.

When approaching vendors, care should be taken not to over- or under- specify requirements. For more information, visit the Project Management for IoT page. It may also be beneficial to invite shortlisted vendors to site to conduct data discovery tasks or solve smaller problems that will help determine if they will be a good fit to help the organisation implement a large-scale data analytic solution.

Challenges

Industry uses a small fraction of available data due to siloed data in legacy systems and leaders’ scepticism about the impact of technologies such as IoT.

For example, added value for a commercial building could be derived from integration of available data into building management systems and building intelligence systems to perform energy analytics  and management to improve efficiency, or condition monitoring and predictive maintenance.

Three factors which contribute to the slow uptake of IoT data analytics in industry are:

• multiple data points (including electrical, thermal and mechanical energy, as well as process and production data) which may be housed in separate servers
• proprietary or inflexible data collection and storage solutions which are difficult to integrate
• skills gaps in staff and management in understanding the benefits of data analytics

Types of solutions

There are several different models of IoT data analytics solutions as shown in the diagram below.

Diagram courtesy of Umesh Bhutoria, EnergyTech Ventures

A stand-alone system could involve purchasing metres or sensors and asking a vendor to integrate them. This model has the potential to be influenced by the vendor rather than the user organisation’s requirement and does not provide integration with existing data.

The second model, data as a service, provides monitoring and automated reports, but will not necessarily include integration with legacy data.

Insights as a service is a model that is gaining in popularity, and is applicable to organisations with mature data infrastructure, who know what data is available and how the organisation aims to use it. It is typically a cloud-based service that uses company, user and third-party data to provide insights, as well as offering support in using these insights to meet the goals of the organisation. Existing data is also connected and centralised, as shown in the diagram below.

Diagram courtesy of Umesh Bhutoria, EnergyTech Ventures

The choice of solution should be based on the benefits it will bring to the organisation, weighed against the pre- and post- purchase effort, cost and ease and flexibility of use.

The role of data analytics in IoT

A non-exhaustive list of advanced data analytic applications within IOT applications is listed below. The majority of the applications listed revolve around the broad categories of asset management, planning, and performance management. The IOT has helped businesses to address these applications in a more holistic manner than was previously possible.

• Predictive maintenance
• Energy usage optimisation and energy analytics
• Downtime minimisation
• Network performance management
• Device performance effectiveness
• Load balancing optimisation
• Loss prevention
• Capacity planning
• Asset management and inventory tracking
• Demand forecasting
• Pricing optimisation
• Disaster planning and recovery
• Yield management

Sources: The information on this page has been sourced primarily from the following:

• Webinar titled The data management perspective on IoT by Arthur Baoustanos, Managing Director, aib Consulting Services
• Case Study titled Studying movement behaviour in a building: A case study of obtaining analytics from IoT Data
• Webinar titled “The Data Indigestion Crisis: New approaches to Energy Analytics” by Umesh Bhtoria, Founder and CEO, EnergyTech Ventures