Any questions? Feel completely free to contact us

Utilities And Energy Industry

Predictive Analytics and the Utilities Sector

The power industry has grown considerably from its humble 19th century origins. From the first electric bulb in the 1870s to the nuclear power plants of the 21st century, the technology used to efficiently create, harness, and deliver power has evolved at a breakneck speed. Yet no matter how advanced the technology, there are always factors beyond our control, such as weather patterns, maintenance issues, overburdened electric systems, and measurement inefficiencies. Understanding the underlying data behind these issues, and how to predict them, is key to developing a successful predictive analytics strategy for the power industry.

Engaging with Customers in the Power Industry

In today’s age, providing a reliable electrical current is a given and, more often than not, is taken for granted — customers simply expect that their lights will always come on and their appliances will work. The topic of customer service in the power industry (excluding billing) is seldom a critical issue until something goes wrong, at which point customers quickly become irate… even if the power interruption is relatively short. In the power industry, knowing when problems are most likely to happen is key to customer engagement.

Data science plays a critical role in transforming raw production & performance data into power service interruptions and energy consumption metrics. Predictive analytics solutions, such as Dataiku Data Science Studio (DSS), enable users - from data scientists, to data analysts, to data ops - to make sense out of the raw data and using machine learning algorithms to understand and anticipate future events & trends.

Using Dataiku DSS in the Power Industry

Data Science Studio (DSS) is a powerful platform that enables companies to take advantage of raw data by transforming it into highly useful analytical forecasts. This raw data frequently includes critically useful information that is left undiscovered because its extraction is oten too costly and burdensome. Dataiku DSS facilitates this data exploration via a 3-step process:

  • Collection: DSS can collect data from any source, from SQL databases to simple Excel spreadsheets;
  • Cleansing: Raw data is often unformatted, unparsed, and contains missing values. The cleansing process homogenizes the format & structure of the data and removes extraneous data, effectively levelling the dataset playing field;
  • Analytics: Dataiku DSS uses powerful machine learning algorithms to create models designed to predict future events. You are in complete control of which datasets to include for analysis. For example, regional household electric usage data could be combined with weather patterns (heat/cold indexes) to determine the impact of seasonal changes on power usage. These models could be tweaked, as needed, to predict future power consumption based on forecasted weather data. The results of this analysis could be used for both public dissemination (e.g., sharing valuable data with consumer / energy advocacy groups) and/or corporate use (e.g., load testing on supply systems given predicted demand).

Developing Smarter Technology

Accurate real-time usage data, such as data transmitted from smart meters, empowers utility companies to build reliable predictive analytics solutions. Smart meters have changed the way utility companies interact with their customers, effectively enabling both parties to benefit from highly accurate data. Customers are billed for exact consumption and can make their own energy saving decisions, while utility companies are immediately provided with valuable real-time usage data.

This data can be fed into Dataiku Data Science Studio and combined with any other dimensional data to build models based on advanced machine-learning algorithms. For example, a Dataiku DSS-created app could score customers based on expected usage based on historical, seasonal, and trend data. This information, when geolocalized & visualized, would enable utility companies to quickly forecast usage for specific regional areas. Companies could then engage directly with those customers in the event of forecasted outages or perhaps offer targeted incentives for decreased usage.

Success Stories