Future Load Prediction with Neural Network Time Series Prediction
3 views (last 30 days)
Show older comments
Hello,First of All I would like to explain my project.I have 5 years daily and hourly load data.Also I have another column indicates day type.I need to get next 5 days prediction of load values.When I tried to predict this,i saw that I should use NARX network Time series. So My dataset is like that;
913 days-Hour1,Hour2,....Hour24,Daytype,
and My targetset;
913 days-Hour1.Hour2,...,Hour24
My first Question:May dataset and targetset include different number of days? Second Question:When I run the network,I have 911 days prediction,however I have 913 days.When I searched in answer i saw that the reason is number of delays.How will I solve this problem?
Last Question:After the network is trained ,how i will get next 5 days prediction?I want to write last 5 days load data to get next 5 days load data.Could anyone help me please?I really need this.Thanks for your interests!
0 Comments
Accepted Answer
Greg Heath
on 8 Aug 2014
load data: What does that mean? load of what?
day type: what does that mean? hot/warm/cold? windy/rainy/snowy? ;
If you have one series use narnet because you do not have an exogeneous input.
Find the statistically significant delays by using the autocorrelation function
Do not use the default dividerand. I recommend either divideblock or divideind
I have covered quite a bit in previous posts. Search the NEWSGROUP and ANSWERS using
greg narnet
Hope this helps.
Greg
Thank you for formally accepting my answer
Greg
0 Comments
More Answers (1)
See Also
Categories
Find more on Sequence and Numeric Feature Data Workflows in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!