LLP

Introducción

Para el análisis de frecuencias y patrones en los errores en el aprendizaje del CPDnA se leen los datos del fichero LLP.csv proporcionado por Javier Villalva Díez.

#
setwd('~/git/JVD_HEALTH/JVD_LLP')
dd=read.csv(file="LLP3.csv",sep=",",skip=2,header=FALSE,stringsAsFactors=FALSE)
dd$row=as.numeric(rownames(dd))
colnames(dd)=c("Country","ID","Time","CW","PO","KPI",paste("Step_",1:(ncol(dd)-7),sep=""),"Row")
dd$Time=as.numeric(sub('CW','',dd$Time))

Vamos a agregar el rendimiento de cada PO

#
resume_ciclo=function(x) {
  res=ldply(x,function(u){
    dlT=which(u$Nerr==0)[1]
    lm1=lm(Nerr~ 1+CW,data=u[1:dlT,])
    return(data.frame(
      ID = u$ID[1],
      minT=u$Time[1],
      extT=nrow(u),
      dlT =which(u$Nerr==0)[1],
      nTE = sum(u$Nerr),
      FEL = u$Nerr[1],
      numE=length(which(u$Nerr==0)),
      i0  = coef(lm1)[1],
      i1  = coef(lm1)[2],
      iniK =u$KPI[1],
      avgKl=mean(u$KPI[which(u$Nerr==0)],na.rm=TRUE),
      sdKl=sd(u$KPI[which(u$Nerr==0)],na.rm=TRUE) ))
  })
  names(res)[which(names(res)==".id")]="idpo"
  return(res)
}
#
pintaPO=function(matcpdna,lppo,name=NULL){
  mint=min(matcpdna$minT)
  maxt=max(matcpdna$minT+matcpdna$extT-1)
  datp=as.data.frame(matrix(NA,ncol=(nrow(matcpdna)+1),nrow=(maxt-mint+1)))
  colnames(datp)=c("CW",rownames(matcpdna))
  datp[,1]=mint:maxt
  for(i in 1:nrow(matcpdna)) {
    j1=which(datp[,1]==matcpdna[i,"minT"])
    j2=which(datp[,1]==(matcpdna[i,"extT"]+matcpdna[i,"minT"]-1))
    datp[j1:j2,(i+1)] = lppo[[i]]$Nerr
  }
  datpm=melt(datp,.(CW))
  colnames(datpm)[2]="CPDnA"
  p = ggplot(data=datpm,aes(x=CW,y=value,group=CPDnA, colour=CPDnA)) + 
      geom_line() + geom_point() + theme_bw() +  ylab("Num. of Errors")
  print(p)
}
#
dd2 = dd[,c(1:6,ncol(dd))]
dd2$Nerr = apply(dd[,7:(ncol(dd)-1)],1,sum,na.rm=T)
ddl=split(dd2,f=dd2[,"ID"])
#
lpo=list()
idx=ldply(ddl,function(x){return(x$PO[1])})
for (i in 1:max(dd2$PO)) {
  lpo[[i]]=ddl[idx[idx[,2]==i,1]]
  names(lpo[[i]])=1:length(lpo[[i]])
  names(lpo)[i]=paste("PO",i,sep="")
}
#
respo=lapply(lpo,resume_ciclo)
for(i in 1:length(respo)) {
  pintaPO(respo[[i]],lpo[[i]])
}

#

Representación de los patrones de aprendizaje

#
print(xtable(ldply(respo)),type="html")

	.id	idpo	ID	minT	extT	dlT	nTE	FEL	numE	i0	i1	iniK	avgKl	sdKl
1	PO1	1	1	1.00	13	9	44	10	5	10.81	-1.18	100	32.40	2.07
2	PO1	2	2	10.00	8	4	12	5	5	7.50	-1.80	100	69.20	9.20
3	PO1	3	3	9.00	3	3	6	5	1	7.00	-2.50	100	92.00
4	PO2	1	4	1.00	21	17	86	10	5	10.44	-0.60	107	52.40	2.97
5	PO2	2	5	10.00	12	8	31	8	5	8.54	-1.04	100	56.40	3.65
6	PO2	3	6	13.00	9	5	18	6	5	8.10	-1.50	100	71.40	6.54
7	PO2	4	7	17.00	8	4	8	5	5	6.00	-1.60	100	83.60	3.58
8	PO3	1	8	1.00	11	6	40	10	5	11.80	-1.94	100	95.00	9.35
9	PO3	2	9	4.00	14	10	39	8	5	7.80	-0.71	89	54.60	3.36
10	PO3	3	10	10.00	9	5	23	10	5	11.20	-2.20	100	60.20	1.92
11	PO3	4	11	14.00	7	3	6	5	5	7.00	-2.50	100	90.60	7.06
12	PO4	1	12	1.00	14	10	47	10	5	10.33	-1.02	100	68.40	2.70
13	PO4	2	13	4.00	10	6	29	9	5	11.13	-1.80	100	74.20	3.27
14	PO4	3	14	10.00	14	10	48	10	5	10.20	-0.98	100	94.00	3.74
15	PO4	4	15	14.00	9	5	15	5	5	7.20	-1.40	89	84.00	1.58

#