We have evaluated a wide range of state-of-the-art
methods and strategies for short-term forecasting of household
electricity consumption, which is a key capability
in many smart grid applications. Although our current
data base is limited, we were able to gain useful insights
into their performance at dierent levels of granularity
and forecasting horizon length. We showed that
without further renement of advanced methods such
as ARIMA and neural networks, the persistence forecasts
are hard to beat in most situations. Especially
in households with demand proles that remain constant
for many hours during a typical day, advanced
forecasting methods provide little value, if they are not
embedded into a framework that adapts their use to individual
household attributes. Future work will focus
on the design of such frameworks and evaluate them
based on representative data.