Skip to contents

DTW-based clustering

Source: R/ops-dtw.R
dtwClustering.Rd

Clusters waveforms using DTW distance with hierarchical or k-medoids method.

Usage

dtwClustering(
  x,
  k = 2,
  method = c("hierarchical", "kmedoids"),
  linkage = "ward.D2",
  window_size = NULL
)

Arguments

x

A PhysioExperiment object or matrix (time x observations).

k

Number of clusters.

method

Clustering method: "hierarchical" or "kmedoids".

linkage

For hierarchical: "ward.D2", "complete", "average", etc.

window_size

Sakoe-Chiba band width.

Value

A list containing:

clusters

Cluster assignments

centers

Cluster centers (medoids or DBA averages)

distance_matrix

DTW distance matrix used

References

Sakoe H, Chiba S (1978). "Dynamic programming algorithm optimization for spoken word recognition." IEEE Transactions on Acoustics, Speech, and Signal Processing, 26(1), 43-49.

See also

dtwDistance(), dtwDistanceMatrix(), dtwAverage()

Examples

# Cluster gait patterns
set.seed(123)
t <- seq(0, 100, length.out = 100)

# Generate two groups with different patterns
group1 <- sapply(1:15, function(i) {
  sin(2 * pi * t / 100) * 30 + rnorm(100, 0, 3)
})
group2 <- sapply(1:15, function(i) {
  sin(2 * pi * t / 100 + pi/4) * 20 + rnorm(100, 0, 3)
})
data <- cbind(group1, group2)

result <- dtwClustering(data, k = 2)
table(result$clusters)
#> 
#>  1  2 
#> 15 15