The MCM model accurately forecasts electricity use in buildings


In a new study published in the peer-reviewed journal Applied Energy, the forecasting model MCM has been proven to be a simple but accurate alternative to other so-called benchmark models. The model has been developed by Joakim Munkhammar at the Division of Civil Engineering and Built Environment.

The MCM (Markov-chain mixture distribution) model was used in the study to make short-term forecasts of electricity use in residential buildings. The MCM model proved to be performing similarly to the Quantile Regression method, and much better than the existing benchmark model Persistence Ensemble. Due to its relative simplicity, in combination with its accuracy in electricity use forecasting, the MCM model is recommended for use as a benchmark model against which to compare other, more advanced forecasting models.  

The paper is published as Open Access and can be read in its entirety here.


Last modified: 2022-02-08