Temporalization (5.0.2)
For time series data, this new tool allows creating Temporal Clones of selected (or all) variables.
A Temporal Clone has the same discretization as the original node (time step 0).
New Classes are created for each time slice.
Temporal Indexes are associated with each variable. These indexes automatically define forbidden arcs, such that there can be no connection from the "future to the past."
Unsupervised Learning with 10 Temporal Clones
Supervised Learning with 10 Temporal Clones
By using the Markov Blanket Learning algorithm, we find here that the order of the Markovian Model for this problem is 5
