GMN Example

A simple example demonstrates the processing pipeline on a toy data set. The network consists of 5 nodes: A B C D Out. Each node represents a time series of length 1000 points with network structure and time series:

---

Interaction Matrix

The interaction matrix defines the GMN network created with the InteractionMatrix.py application (see Interaction Matrix). Help can be shown with the -h argument. We create the interaction matrix from data file TestData_ABCD.csv using the EDM convergent cross mapping (CCM) metric, storing the output interaction matrix in ABCD_iMatrix_E5_tau-3_CCM.csv. CCM is passed an embedding dimension of E=5, and time delay of tau=-3.

./apps/InteractionMatrix.py -d ./data/TestData_ABCD.csv -oc ./output/ABCD_iMatrix_E5_tau-3 -ccm -E 5 -t -3 -P

Network Creation

The CreateNetwork.py application (see Create Network) reads the interaction matrix and creates the networkx directed graph object, here stored in a binary file using the python pickle module.

./apps/CreateNetwork.py -i ./output/ABCD_iMatrix_E5_tau-3_CCM.csv -t Out -o ./output/ABCD_Network_E5_tau-3_CCM.pkl -d 4 -P -l spring

---

Generative Mode

With a GMN network we can run GMN in generative mode according to the parameters specified in a configuration file (see Parameters). [EDM] parameters are defined in EDM Parameters.

Define the configuration file ./network/ABCD_Out.cfg as :

[GMN]
mode             = Generate
predictionStart  = 700
predictionLength = 300
backend          = serial
kernel           = True
outPath          = ../output
dataOutFile      =
showPlot         = True
plotType         = state
plotColumns      = Out A B C D
plotFile         =

[Network]
name       = ABCD 4 Driver
targetNode = Out
file       = ./network/ABCD_Test/ABCD_Network_E3_T0_tau-1_CMI.pkl
data       = ./data/TestData_ABCD.csv

[Node]
info       = EDM Simplex Manifolds
function   = Simplex

[EDM]
E        = 7
Tp       = 1
tau      = -3
validLib = 

[Scale]
factor = 1
offset = 0

From the python console import the gmn package, create the GMN object and run the network in generative mode:

import gmn

G = gmn.GMN( configFile = './config/ABCD_Out.cfg' )

G.Generate()

G.DataOut.tail( 5 )
     Time       A       C       D         B       Out
295   996 -0.2487 -0.5018  0.7500  0.985236 -0.979370
296   997 -0.1874 -0.4708  0.7937  0.985842 -0.991504
297   998 -0.1253 -0.4248  0.8177  0.965066 -0.973041
298   999 -0.0628 -0.3671  0.8224  0.923630 -0.931681
299  1000  0.0000 -0.3016  0.8090  0.862222 -0.871642

The output state plot shows the library (observed time series & state-space) in blue, and GMN generated values in orange.