search

Waikato / wekaDeeplearning4j

Weka package for the Deeplearning4j java library
https://deeplearning.cms.waikato.ac.nz/
GNU General Public License v3.0
184 stars 197 forks source link

Problem evaluating forecaster #36

Closed lumio83 closed 6 years ago

lumio83 commented 6 years ago

When using wekadeeplearning4j for time series forecast, and setting up LSTM layer and Rnnoutputlayer I get this message: Problem evaluating forecaster indexes must be same length as array rank.

What could be the problem here ?

braun-steven commented 6 years ago

Hi lumio83,

which class are you using? RnnSequenceClassifier? Are you using the Weka GUI or did you write a java program? If it is the second, could you provide your code snippet and the stack trace? If you are using the GUI, could you provide the configuration of your setup?

lumio83 commented 6 years ago

Hi I am using Weka GUI with dl4jmlp classifier for time series forecast. In layers just one LSTM layer + rnnoutput layer.

braun-steven commented 6 years ago

Time series forecasting is currently not supported but rather WIP here. The RnnForecaster (when ready) will be what you are looking for.