Demand Forecast with Azure AI - 10 weeks PoC

Informatika a.d.

Build as most accurate as possible electricity demand forecasting model by using Azure Machine Learning

The solution provides support to demand forecasters to utilize power of Azure advanced analytics to accurately predict needs for electrical energy of electricity market in short and mid-term. Having deep machine learning insights (Azure ML) into their historical consumption data, demand forecasters are able to collaborate with production and trading team to most profitably manage shortages or surpluses of their energy assets. Informatika's solution is using following Azure Services: Azure Machine Learning, Azure Data Factory, Azure Batch, Azure Logic Apps and Azure SQL Database. We learn and analyze customer processes and get familiar with different departments who are involved in the business process like dispatchers, brokers, traders, and got approval from IT and Security to use Azure. Afterward, we deliver two days workshop and build a basic model "one day ahead", based on the customer's historical data. The data we need from the customer side is "Dates with accomplished consumption "and "Daily temperature ". Offer a solution for everyone (operations, management, teams outside trading for transparency). In the end, we are building the PoC model for hourly consumption. If possible, we will involve more ponders such as Sunrise, Sunset, number of Sunny hours, Min, Max, and Average temperature, Humidity, Cloudiness, Wind.