I am venturing into a new field of electricity load forecasting. Most of the papers I read are either based on neural network, support vector regression or multiple linear regression.
I am wondering whether there are other methods simpler and same as or more superior that the machine learning methods.
What is the current trend. Are a lot of people, company using Machine Learning
Hi. I have experimented with many of these techniques. Exponential smoothing methods found good amoung statistical methods.
Please refer my paper on Double Exponential smoothing.
A simple method is to store input/output pairs in a list.
If you want to know the output for a new input, search
in the list for a pair whose input best matches the
new input, and put out its output.
Later you can improve this with statistics that
provides a measure about how useful a stored pair
was, and remove the badly performing ones.
Regards,
Joachim
Not all the time neural network, support vector regression or multiple linear regression methods are used to forecasting. You may follow my paper where I have used some other machine learning algorithms to predict weather.
https://arxiv.org/abs/1406.4756
http://ieeexplore.ieee.org/document/7344438/
Article A Review on Application of Data Mining Techniques to Combat ...
- Badges
- Science topic