Tools | Time Series
Updated Feature: Network Temporalization
If a dataset describes time series, Network Temporalization can directly create Temporal Clones of the selected variables.
Prior to the 5.4 release, the clones were created as lead variables, denoted by Node_t, Node_t+1, ..., Node_t+n.
Now, Network Temporalization creates clones as lagged variables, denoted by Node_t, Node_t-1, ..., Node_t-n. An example is shown below.
Updated Feature: Forecasting | Prediction Mode
The Prediction Mode now offers two options for forecasting:
- Predicting the Modal Value, i.e. the value of the state that has the highest posterior probability.
- Predicting the Expected Value: the Expected Value is computed from the entire posterior probability distribution.
Updated Feature: Forecasting | Output
You now have the option of saving the forecasted values as an internal Test Database, i.e. as a test set, or to save them as a new file.