The paper describes a method for predicting energy con-
sumption. For this purpose, several machine learning algo-
rithms are compared in the paper. Chosen as the best algorithm
was the local linear regression. The main contribution of the
paper lies in comparing different machine learning algorithms
and in evaluating the accuracy for different lengths of time
intervals. Another benefit is the achieved accuracy, which
exceeds that in comparable works.
As is evident from the results, it is very difficult to accu-
rately capture an exact pattern of electricity consumption. One
of the reasons why this is so, is the considerable influence of
weather and other factors such as the economy, population
growth, etc. Future work will therefore consider looking for
trends that affect consumption and can be reliably predicted.