criterion: ~ Function: criterion ~

Description

Given a LongData and a Partition, the fonction criterion calculate some criterions that are classicaly use to estimate the quality of a clusterization.

Usage

criterion(object, partition, method)

Arguments

object

[LongData]: object on which the criterion are calculate

partition

[Paritition]: clusterization of the LongData

method

[character]: if some value are missing in the LongData, it is necessary to impute them. The function criterion call the function imputation using the

Value

A list:
- varBetween
{[matrix]: variance between}
varWithin[matrix]: variance within
calinski[numeric]: Calinski and Harabatz criterion

Details

Given a LongData and a Partition, the fonction criterion calculate some criterions that are classicaly use to estimate the quality of a clusterization. If some individual have no clusters (ie if Partition has some missing values), the corresponding trajectories are exclude from the calculation. Note that if there is an empty cluster or an empty longData, most of the criterions are anavailable.

Examples

Run this code

##################
### Preparation of some artificial data
par(ask=TRUE)
traj <- gald()

### Correct partition
part1 <- partition(rep(1:4,each=50),4)
(cr1 <- criterion(traj,part1))
plot(traj,part1,main=paste("Calinski =",formatC(cr1[["calinski"]])))


### Random partition
part2 <- partition(floor(runif(200,1,5)),4)
(cr2 <- criterion(traj,part2))
plot(traj,part2,main=paste("Calinski =",formatC(cr2[["calinski"]])))

### Partition with 3 clusters instead of 4
part3 <- partition(rep(c(1,2,3,3),each=50),3)
(cr3 <- criterion(traj,part3))
plot(traj,part3,main=paste("Calinski =",formatC(cr3[["calinski"]])))

### Comparisons of the Partition
cr1["calinski"]
cr2["calinski"]
cr3["calinski"]
par(ask=FALSE)