ramidst (version 0.1.0)

nb_concept_drift_detector_from_stream: Naive Bayes concept drift detector from an AMIDST data stream

Description

Naive Bayes concept drift detector from an AMIDST data stream

Usage

nb_concept_drift_detector_from_stream(input_stream, class_index = -1L, window_size, transition_variance = 0.1, hidden_vars = 1L)

Arguments

input_stream

an AMIDST input stream

class_index

the index of the class variable in the list of variables

window_size

the number of items in the stream to be analysed simultaneously

transition_variance

the variance of the transition distribution

hidden_vars

the number of global hidden variables to include in the model

Value

the value of the hidden variables for each window

References

H. Borchani, A.M. Martinez, A.R. Masegosa, H. Langseth, T.D. Nielsen, A. Salmeron, A. Fernandez, A.L. Madsen, R.Saez (2015) Modeling concept drift: A probabilistic graphical model based approach. IDA'2015. Lecture Notes in Computer Science 9385, 72-83.

Examples

## Not run: 
# data <- amidst_data_stream(system.file("extdata","sea.arff",
# package="ramidst"))
# results <- nb_concept_drift_detector_from_stream(data,class_index = -1L,
# window_size=1000L,transition_variance=0.1,hidden_vars=1L)
# re <- 0
# for (k in 1:length(results)) re[k] <- results[[k]]
# ymin = min(re)-0.05
# ymax = max(re)+0.05
# plot(re,type="l",ylim=c(ymin,ymax),ylab="Hidden variable",
# xlab="Instance number (x 1000)")
# abline(v=15,col="red")
# abline(v=30,col="red")
# abline(v=45,col="red")
# ## End(Not run)