hcd_eddm {harbinger} | R Documentation |
EDDM (Early Drift Detection Method) aims to improve the detection rate of gradual concept drift in DDM, while keeping a good performance against abrupt concept drift. doi:2747577a61c70bc3874380130615e15aff76339e
hcd_eddm(
min_instances = 30,
min_num_errors = 1,
warning_level = 0.95,
out_control_level = 0.9
)
min_instances |
The minimum number of instances before detecting change |
min_num_errors |
The minimum number of errors before detecting change |
warning_level |
Necessary level for warning zone |
out_control_level |
Necessary level for a positive drift detection |
hcp_eddm
object
library("daltoolbox")
library("ggplot2")
set.seed(6)
# Loading the example database
data(har_examples)
#Using example 1
dataset <- har_examples$example1
cut_index <- 60
srange <- cut_index:row.names(dataset)[nrow(dataset)]
drift_size <- nrow(dataset[srange,])
dataset[srange, 'serie'] <- dataset[srange, 'serie'] + rnorm(drift_size, mean=0, sd=0.5)
head(dataset)
plot(x=row.names(dataset), y=dataset$serie, type='l')
# Setting up time series regression model
model <- hanct_kmeans()
# Fitting the model
model <- fit(model, dataset$serie)
# Making detection using hanr_ml
detection <- detect(model, dataset$serie)
# Filtering detected events
print(detection[(detection$event),])
# Drift test
drift_evaluation <- data.frame(!(detection$event == dataset$event)) * 1
model <- fit(hcd_eddm(min_instances=10), drift_evaluation)
detection_drift <- detect(model, drift_evaluation)
grf <- har_plot(model, dataset$serie, detection_drift)
grf <- grf + ylab("value")
grf <- grf
plot(grf)