class KMeansClustererPP extends KMeansClustererHW
The KMeansClustererPP
class cluster several vectors/points using
the Hartigan-Wong algorithm.
- Alphabetic
- By Inheritance
- KMeansClustererPP
- KMeansClustererHW
- KMeansClusterer
- Error
- Clusterer
- AnyRef
- Any
- Hide All
- Show All
- Public
- All
Instance Constructors
-
new
KMeansClustererPP(x: MatriD, k: Int, flags: Array[Boolean] = Array (false, false))
- x
the vectors/points to be clustered stored as rows of a matrix
- k
the number of clusters to make
- flags
the flags used to adjust the algorithm
Value Members
-
def
calcCentroids(x: MatriD, to_c: Array[Int], sz: VectoI, cent: MatriD): Unit
Calculate the centroids based on current assignment of points to clusters and update the 'cent' matrix that stores the centroids in its rows.
Calculate the centroids based on current assignment of points to clusters and update the 'cent' matrix that stores the centroids in its rows.
- x
the data matrix holding the points {x_i = x(i)} in its rows
- to_c
the cluster assignment array
- sz
the sizes of the clusters (number of points)
- cent
the matrix holding the centroids in its rows
- Definition Classes
- Clusterer
-
def
centroids: MatriD
Return the centroids.
Return the centroids. Should only be called after
train
.- Definition Classes
- KMeansClusterer → Clusterer
-
def
checkOpt(x: MatriD, to_c: Array[Int], opt: Double): Boolean
Check to see if the sum of squared errors is optimum.
Check to see if the sum of squared errors is optimum.
- x
the data matrix holding the points
- to_c
the cluster assignments
- opt
the known (from human/oracle) optimum
- Definition Classes
- Clusterer
-
def
classify(z: VectoD): Int
Given a new point/vector 'z', determine which cluster it belongs to, i.e., the cluster whose centroid it is closest to.
Given a new point/vector 'z', determine which cluster it belongs to, i.e., the cluster whose centroid it is closest to.
- z
the vector to classify
- Definition Classes
- KMeansClusterer → Clusterer
-
def
cluster: Array[Int]
Return the cluster assignment vector.
Return the cluster assignment vector. Should only be called after
train
.- Definition Classes
- KMeansClusterer → Clusterer
-
def
csize: VectoI
Return the sizes of the centroids.
Return the sizes of the centroids. Should only be called after
train
.- Definition Classes
- KMeansClusterer → Clusterer
-
def
distance(u: VectoD, cn: MatriD, kc_: Int = -1): VectoD
Compute the distances between vector/point 'u' and the points stored as rows in matrix 'cn'
Compute the distances between vector/point 'u' and the points stored as rows in matrix 'cn'
- u
the given vector/point (u = x_i)
- cn
the matrix holding several centroids
- kc_
the number of centroids so far
- Definition Classes
- Clusterer
-
def
distance2(u: VectoD, cent: MatriD, cc: Int): VectoD
Compute the adjusted distance to point 'u' according to the R2 value described in the Hartigan-Wong algorithm.
Compute the adjusted distance to point 'u' according to the R2 value described in the Hartigan-Wong algorithm.
- u
the point in question
- cent
the matrix holding the centroids
- cc
the current cluster for point u
- Definition Classes
- KMeansClustererHW
-
val
flags: Array[Boolean]
- Definition Classes
- KMeansClusterer
-
final
def
flaw(method: String, message: String): Unit
- Definition Classes
- Error
-
def
initCentroids(): Boolean
Initialize the centroids according to the k-means++ technique.
Initialize the centroids according to the k-means++ technique.
- Definition Classes
- KMeansClustererPP → Clusterer
-
def
name(c: Int): String
Return the name of the 'c'-th cluster.
-
def
name_(nm: Strings): Unit
Set the names for the clusters.
-
def
setStream(s: Int): Unit
Set the random stream to 's'.
Set the random stream to 's'. Method must be called in implemeting classes before creating any random generators.
- s
the new value for the random number stream
- Definition Classes
- Clusterer
-
def
show(l: Int): Unit
Show the state of the algorithm at iteration 'l'.
Show the state of the algorithm at iteration 'l'.
- l
the current iteration
- Definition Classes
- KMeansClusterer
-
def
sse(x: MatriD, c: Int, to_c: Array[Int]): Double
Compute the sum of squared errors from the points in cluster 'c' to the cluster's centroid.
Compute the sum of squared errors from the points in cluster 'c' to the cluster's centroid.
- x
the data matrix holding the points
- c
the current cluster
- to_c
the cluster assignments
- Definition Classes
- Clusterer
-
def
sse(x: MatriD, to_c: Array[Int]): Double
Compute the sum of squared errors within all clusters, where error is indicated by e.g., the distance from a point to its centroid.
Compute the sum of squared errors within all clusters, where error is indicated by e.g., the distance from a point to its centroid.
- x
the data matrix holding the points
- to_c
the cluster assignments
- Definition Classes
- Clusterer
-
def
sst(x: MatriD): Double
Compute the sum of squares total for all the points from the mean.
Compute the sum of squares total for all the points from the mean.
- x
the data matrix holding the points
- Definition Classes
- Clusterer
-
def
train(): KMeansClusterer
Iteratively recompute clusters until the assignment of points does not change.
Iteratively recompute clusters until the assignment of points does not change. Initialize by randomly assigning points to 'k' clusters.
- Definition Classes
- KMeansClusterer → Clusterer
-
def
update_pmf(c: Int): Discrete
Update the probability mass function (pmf) used for picking the next centroid.
Update the probability mass function (pmf) used for picking the next centroid. The farther 'x_i' is from any existing centroid, the higher its probability. Return the corresponding distance-derived random variate generator.
- c
the current centroid index