A class that generates random data based on RandomRBFGeneratorEvents implemented in MOA.
DSD_RandomRBFGeneratorEvents(k = 3, d = 2, numClusterRange = 3L,
kernelRadius = 0.07, kernelRadiusRange = 0, densityRange = 0,
speed =100L, speedRange = 0L, noiseLevel = 0.1,
noiseInCluster = FALSE, eventFrequency = 30000L,
eventMergeSplitOption = FALSE, eventDeleteCreate = FALSE,
modelSeed = NULL, instanceSeed = NULL)
The average number of centroids in the model.
The dimensionality of the data.
Range for numner of clusters.
The average radius of the micro-clusters.
Deviation of the number of centroids in the model.
Density range.
Kernels move a predefined distance of 0.01 every X points.
Speed/Velocity point offset.
Noise level.
Allow noise to be placed within a cluster.
Frequency of events.
Merge and split?
Delete and create?
Random seed for the model.
Random seed for the instances.
An object of class DSD_RandomRBFGeneratorEvent
(subclass of DSD_MOA
, DSD
).
There are an assortment of parameters available for the underlying MOA data
structure, however, we have currently limited the available parameters to
the arguments above. Currently the modelSeed and instanceSeed are set to
default values every time a DSD_MOA
is created, therefore the
generated data will be the same. Because of this, it is important to set
the seed manually when different data is needed.
The default behavior is to create a data stream with 3 clusters and concept drift. The locations of the clusters will change slightly, and they will merge with one another as time progresses.
MOA: Massive Online Analysis, a Framework for Stream Classification and Clustering Albert Bifet, Geoff Holmes, Bernhard Pfahringer, Philipp Kranen, Hardy Kremer, Timm Jansen, Thomas Seidl. Journal of Machine Learning Research (JMLR).
# NOT RUN {
stream <- DSD_RandomRBFGeneratorEvents()
get_points(stream, 10, class=TRUE)
# }
# NOT RUN {
animate_data(stream, n=5000, pointInterval=100, xlim=c(0,1), ylim=c(0,1))
# }
Run the code above in your browser using DataLab