Neural network is commonly used for time series forecasting, however most of the time series has continued data (it means real data or float data). My question is if we have a time series with a certain set of values can we use neural network for prediction?
for example the set of values is: {1,2,3,4,5}
and time series is:
2,2,2,1,5,1,4,3,5,2,1,1,2,5,4,3,4,5,1,2,5,5,5,1,1,2,5,3,3,3,?
and we want to predict the next value.
ANN are used for classification and regression, in this case Time Series Forecasting is a regression problem. But this particular scenario could be viewed as a classification problem.
In any case, you'll need a larger data set and define some elements in order to setup an ANN application: ANN scheme, variable inputs (lagged values maybe), training algorithm, error measure, training set and test set (supervised learning).
The link published by Mahboobeh Parsapoor shows an example of forecasting time series with lagged variables: forecast (t) using (t-1),(t-2),..,(t-N)
Best
Dear Ali,
Here are a few good articles you may refer to:
Adhikari, R. (2015). A neural network based linear ensemble framework for time series forecasting. Neurocomputing, 157, 231-242.
Frank, R. J., Davey, N., & Hunt, S. P. (2001). Time series prediction and neural networks. Journal of intelligent and robotic systems, 31(1-3), 91-103.
Faraway, J., & Chatfield, C. (1998). Time series forecasting with neural networks: a comparative study using the airline data. Applied statistics, 231-250.
http://uhra.herts.ac.uk/bitstream/handle/2299/1565/901700.pdf?sequence=1
http://is.muni.cz/el/1456/jaro2009/PMAPEM/NotSortedYet/Time_series_forecasting_with_neural_networks__Faraway___Chatfield.pdf
https://cld.pt/dl/download/5cf0dc5f-72b0-4b05-953e-484f624b49f6/MyPapers/l984jnf_5843.pdf
Time series data set is different from discrete data set and some learning methods are better for time series and sequential data set such as rnn.
- Badges
- Science topic