HUPI – Energy Consumption Prediction and Behavioural Modelling of an Entity
HUPI
Sector: Industry
Business Case
Automatic energy demand prediction for the next period of time (12 months?), with a precise level of granularity: it predicts the level of energy consumption that each specific customer or building will need in the next day/week/month.
Objectives
Estimate the energy bill and make decisions based on this data. Understand how and why we consume a certain amount in order to change our habits without affecting our productivity or the quality of our lives. Reduce energy consumption and improve sustainability.
Use case
The algorithm begins by using HISTORICAL consumption data, including customer profiles and their consumption habits, and considers the forecasted growth or decline of the network portfolio. It also takes into account the building infrastructure’s characterisation and location. Additionally, it incorporates OPEN data such as meteorological information, specific events, and macroeconomic indicators to predict the potential impacts of seasonality, events, or global situations on production. This process starts with an intelligent multidimensional segmentation, followed by the application of a specific forecasting model for each segment.
Infrastructure
Cloud
Technology
Machine learning and deep learning
Data
The algorithm begins by using HISTORICAL consumption data, including customer profiles and their consumption habits, and considers the forecasted growth or decline of the network portfolio. It also takes into account the building infrastructure’s characterisation and location. Additionally, it incorporates OPEN data such as meteorological information, specific events, and macroeconomic indicators to predict the potential impacts of seasonality, events, or global situations on production. This process starts with an intelligent multidimensional segmentation, followed by the application of a specific forecasting model for each segment.
Resources
Human resources: multidisciplinary team: 1 technical director, 2 mathematicians & data scientists, 1 full stack developer, 1 frontend developer and digital UX expert team, domain expert team. Hupi Elastic Cloud Platform
Difficulties and learning
The definition and variability of external factors and other macro and microeconomic indicators. Team management: coordination between technological experts and domain experts.
KPIs (business impact and metrics of the model)
Estimación del consumo energético del próximo mes por cada perfil. Mejora de la planificación interna.
Funding
Private. The customer paid for the development of this application.
Collaborators, Partners
Own resources. It was a product made by our data team.
