FINAL2

LIBRARY

library(rio)
library(magrittr)
library(polycor)
library(psych)

## 
## Adjuntando el paquete: 'psych'

## The following object is masked from 'package:polycor':
## 
##     polyserial

library(matrixcalc)
library(GPArotation)

## 
## Adjuntando el paquete: 'GPArotation'

## The following objects are masked from 'package:psych':
## 
##     equamax, varimin

library(BBmisc)

## 
## Adjuntando el paquete: 'BBmisc'

## The following object is masked from 'package:base':
## 
##     isFALSE

DATA:

rm(list = ls())
data = import("dataOK_all.xlsx")

## New names:
## • `` -> `...1`

PROCENTAJES:

data$pct_AGUARED = data$agua1_Red / data$agua10_Total
data$pct_RAZONKPC = data$Keiko / data$Castillo
data$tasa_fallecidos <- (data$covidFallecidos / data$covidPositivos) * 1000
data <- data[!(data$provincia %in% c("LIMA")), ]

DATA:

select = c("pct_RAZONKPC", "pct_AGUARED", "tasa_fallecidos")
theData = data [,select]
head(theData,10)

##    pct_RAZONKPC pct_AGUARED tasa_fallecidos
## 1     0.4202271   0.4830430        56.85454
## 2     0.6220444   0.6252053       185.08997
## 3     0.6683045   0.7605543       129.25483
## 4     0.1099285   0.1345204        31.88739
## 5     0.6219261   0.5622011       192.98246
## 6     0.6892180   0.5902965       545.45455
## 7     0.5260536   0.4871039        89.62390
## 8     0.6077419   0.7475528       329.11392
## 9     0.1558544   0.8528790       574.07407
## 10    0.2891608   0.7132928       355.93220

NORMALIZAR:

normalize <- function(x) (x - min(x)) / (max(x) - min(x))
theDataa <- as.data.frame(lapply(theData, normalize))

DATA PARA CLUSTER:

dataClus=theDataa
row.names(dataClus)=data$provincia

#distances
library(cluster)
g.dist = daisy(dataClus, metric="gower")

PAM

library(factoextra)

## Cargando paquete requerido: ggplot2

## 
## Adjuntando el paquete: 'ggplot2'

## The following objects are masked from 'package:psych':
## 
##     %+%, alpha

## Welcome! Want to learn more? See two factoextra-related books at https://goo.gl/ve3WBa

fviz_nbclust(dataClus, pam,diss=g.dist,method = "gap_stat",k.max = 10,verbose = F)

INTERPRETACION PAM:

# pam
set.seed(123)
res.pam=pam(g.dist,k = 3,cluster.only = F)
# nueva columna
dataClus$pam=res.pam$cluster
head(dataClus)

##                      pct_RAZONKPC pct_AGUARED tasa_fallecidos pam
## BAGUA                  0.13562051   0.5595250     0.023039387   1
## BONGARA                0.20722832   0.7252007     0.097345896   2
## CHACHAPOYAS            0.22364210   0.8829363     0.064992009   2
## CONDORCANQUI           0.02552194   0.1533573     0.008572075   1
## LUYA                   0.20718633   0.6517758     0.101919224   2
## RODRÍGUEZ DE MENDOZA   0.23106252   0.6845180     0.306160516   2

#países mal clusterizados
silPAM=data.frame(res.pam$silinfo$widths)
silPAM$provincia=row.names(silPAM)
poorPAM=silPAM[silPAM$sil_width<0,'provincia']%>%sort()
poorPAM

##  [1] "ACOBAMBA"         "ANDAHUAYLAS"      "ATALAYA"          "CANGALLO"        
##  [5] "CANTA"            "CAYLLOMA"         "CHANCHAMAYO"      "CHINCHA"         
##  [9] "FERREÑAFE"        "HUAROCHIRÍ"       "HUAYLAS"          "ICA"             
## [13] "LAMAS"            "LORETO"           "MARISCAL CÁCERES" "OTUZCO"          
## [17] "OYÓN"             "PALPA"            "PICOTA"           "PISCO"           
## [21] "SAN MARCOS"       "SAN MARTÍN"       "SÁNCHEZ CARRIÓN"  "SANTA"           
## [25] "TAYACAJA"         "VILCAS HUAMÁN"

AGNES:

fviz_nbclust(dataClus, hcut,diss=g.dist,method = "gap_stat",k.max = 10,verbose = 
               F,hc_func = "agnes")

res.agnes<- hcut(g.dist, k = 5,hc_func='agnes',hc_method = "ward.D")

dataClus$agnes=res.agnes$cluster

silAGNES=data.frame(res.agnes$silinfo$widths)
silAGNES$provincia=row.names(silAGNES)
poorAGNES=silAGNES[silAGNES$sil_width<0,'provincia']%>%sort()
poorAGNES

##  [1] "CAJABAMBA"         "CAJATAMBO"         "CARAVELÍ"         
##  [4] "CHINCHEROS"        "CORONEL PORTILLO"  "HUÁNUCO"          
##  [7] "HUAYTARÁ"          "JAÉN"              "JAUJA"            
## [10] "LA UNIÓN"          "LEONCIO PRADO"     "LORETO"           
## [13] "LUYA"              "MANU"              "NAZCA"            
## [16] "OCROS"             "OTUZCO"            "OXAPAMPA"         
## [19] "PACHITEA"          "SÁNCHEZ CARRIÓN"   "SANTIAGO DE CHUCO"
## [22] "SUCRE"             "TARATA"            "TARMA"

DIANA

fviz_nbclust(dataClus, hcut,diss=g.dist,method = "gap_stat",k.max = 10,verbose = 
               F,hc_func = "diana")

res.diana <- hcut(g.dist, k = 4,hc_func='diana')
dataClus$diana=res.diana$cluster 

#países mal clusterizados
silDIANA=data.frame(res.diana$silinfo$widths)
silDIANA$provincia=row.names(silDIANA)
poorDIANA=silDIANA[silDIANA$sil_width<0,'provincia']%>%sort()
poorDIANA

## [1] "HUACAYBAMBA"   "LA MAR"        "LEONCIO PRADO" "LORETO"       
## [5] "PASCO"         "PIURA"         "TRUJILLO"      "TUMBES"

PAM

#' Evaluate
fviz_silhouette(res.pam,print.summary = F)

DIANA

fviz_silhouette(res.diana,print.summary = F)

AGNES:

fviz_silhouette(res.agnes,print.summary = F)