Machine Learning: an adoption still linked to a good understanding of earnings

As companies master the basics of reporting, BI, and descriptive analysis , the predictive and prescriptive enters their radar. This is where the value is now unsealed. The problem, especially for companies that market analytics is how to explain their value to customers.

“In some cases, companies understand what we do in the blink of an eye,” says Boris Savkovic, data scientist at BuildingIQ, in an email. “In others, great education efforts are needed. ”

BuildingIQ, located in California, is a company that supports building managers in the control and monitoring of heating and air conditioning infrastructure to improve efficiency and costs. The SaaS solution is based on advanced machine learning algorithms that take into account data on energy consumption, weather forecasts, HVAC (Heating, Ventilation and Air Conditioning), and energy costs. These algorithms, developed with MATLAB (from MathWorks), continuously determine the correct parameterization of HVAC systems. This data is transmitted to the systems via Java code.

The company says its tools can save up to 15% in energy costs. According to Boris Savkovic, most of the work is to ensure that building managers understand the value proposition.

Part of the difficulty is that customers do not interact with Machine Learning’s algorithms. They are not fully developed by the data scientists of BuildingIQ and are hosted on the company’s own servers. The whole works well autonomously from a customer point of view. While this is a win for customers – they do not have to invest in hardware or dedicate resources to development – it makes things a bit abstract.

That’s why it’s important to do the work upstream, he says. His team retrieves customer data and develops models for their buildings, showing them how much they could have saved had they used the right solution in the past. Being able to visualize this data is key. This translates complex predictive analytics into something that trades can understand, even if they do not have advanced data analysis skills. “Presenting visuals and results goes a long way in explaining what BuildingIQ does,” Boris Savkovic continues.

Finally, the change in mentality may well stimulate the adoption of technologies such as Machine Learning’s advanced algorithms. According to him, the construction industry is certainly not the industry’s earliest technology. But this is starting to change and could democratize this type of solution. “The building industry has been very slow to start its digital transformation,” says the manager. “But given the investments made in Smart Cities and Smart Buildings, adoption, as well as market maturity, is expected to increase.

Leave a Reply

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