EvaluaciĆ³n E3
Inca, Santos, Toledo, Ventocilla
26/7/2020
CASO 1: Sistema Educativo del PerĆŗ
Revisamos la informaciĆ³n
educacion=read.csv("https://raw.githubusercontent.com/VictorGuevaraP/Mineria-de-datos-2020/master/Sistema_Educativo_Peru.csv",sep=";")
head(educacion)## region departam almain_1 almapr_2 almase_3 taasi_17 taasp_18 taass_19
## 1 Selva Amazonas 11235 81839 31688 46.0 79.7 55.1
## 2 Sierra Ancash 32254 185875 95346 55.5 87.7 59.4
## 3 Sierra Apurimac 14728 94015 44724 45.3 88.8 63.6
## 4 Costa Arequipa 31601 143318 104735 44.8 94.5 81.3
## 5 Sierra Ayacucho 17173 127033 55667 39.6 89.9 57.7
## 6 Sierra Cajamarca 33304 280254 107485 41.6 90.0 50.0
## taefp_22 taefs_23 doin_24 dopr_25 dose_26 cein_29 cepr_30 cese_31
## 1 77.9 80.5 505 3272 1718 290 1100 199
## 2 79.6 80.9 1627 8460 6930 834 1781 478
## 3 77.4 79.0 609 3711 2277 361 857 207
## 4 90.5 89.3 1969 7534 6869 910 1107 442
## 5 75.1 78.9 816 5585 3502 397 1392 298
## 6 79.1 80.5 1718 11260 6154 894 3532 642summary(educacion)## region departam almain_1 almapr_2
## Length:23 Length:23 Min. : 3298 Min. : 15104
## Class :character Class :character 1st Qu.:11794 1st Qu.: 86892
## Mode :character Mode :character Median :20469 Median :131746
## Mean :21127 Mean :135380
## 3rd Qu.:29979 3rd Qu.:198012
## Max. :41212 Max. :280254
## almase_3 taasi_17 taasp_18 taass_19
## Min. : 9361 Min. :24.90 Min. :79.70 Min. :48.60
## 1st Qu.: 35865 1st Qu.:41.00 1st Qu.:89.15 1st Qu.:57.80
## Median : 57783 Median :45.30 Median :90.40 Median :63.60
## Mean : 68830 Mean :48.09 Mean :90.51 Mean :65.46
## 3rd Qu.:103454 3rd Qu.:55.95 3rd Qu.:93.25 3rd Qu.:74.70
## Max. :141192 Max. :76.00 Max. :95.50 Max. :86.50
## taefp_22 taefs_23 doin_24 dopr_25 dose_26
## Min. :74.10 Min. :77.50 Min. : 157 Min. : 611 Min. : 541
## 1st Qu.:78.50 1st Qu.:80.30 1st Qu.: 559 1st Qu.: 3227 1st Qu.:2020
## Median :84.00 Median :83.10 Median : 927 Median : 5313 Median :3502
## Mean :83.22 Mean :83.26 Mean :1053 Mean : 5569 Mean :4119
## 3rd Qu.:86.90 3rd Qu.:85.30 3rd Qu.:1543 3rd Qu.: 8014 3rd Qu.:6512
## Max. :93.10 Max. :90.90 Max. :2283 Max. :11260 Max. :7907
## cein_29 cepr_30 cese_31
## Min. : 75.0 Min. : 164.0 Min. : 45.0
## 1st Qu.:253.0 1st Qu.: 685.5 1st Qu.:189.5
## Median :437.0 Median :1185.0 Median :264.0
## Mean :489.0 Mean :1252.0 Mean :300.7
## 3rd Qu.:775.5 3rd Qu.:1805.5 3rd Qu.:454.5
## Max. :910.0 Max. :3532.0 Max. :642.0dim(educacion)## [1] 23 16Eliminamos las 2 primeras variables ya que no son numƩricas
educacion_prep=educacion[,3:16]
head(educacion_prep)## almain_1 almapr_2 almase_3 taasi_17 taasp_18 taass_19 taefp_22 taefs_23
## 1 11235 81839 31688 46.0 79.7 55.1 77.9 80.5
## 2 32254 185875 95346 55.5 87.7 59.4 79.6 80.9
## 3 14728 94015 44724 45.3 88.8 63.6 77.4 79.0
## 4 31601 143318 104735 44.8 94.5 81.3 90.5 89.3
## 5 17173 127033 55667 39.6 89.9 57.7 75.1 78.9
## 6 33304 280254 107485 41.6 90.0 50.0 79.1 80.5
## doin_24 dopr_25 dose_26 cein_29 cepr_30 cese_31
## 1 505 3272 1718 290 1100 199
## 2 1627 8460 6930 834 1781 478
## 3 609 3711 2277 361 857 207
## 4 1969 7534 6869 910 1107 442
## 5 816 5585 3502 397 1392 298
## 6 1718 11260 6154 894 3532 642dim(educacion_prep)## [1] 23 14Pruebas preliminares
corrplot(cor(educacion_prep))
chart.Correlation(educacion_prep)
Para la decisiĆ³n respecto a prueba de hipĆ³tesis
P- Valor < nivel de significancia (0.05) Rechazo Ho
cortest(educacion_prep)## Tests of correlation matrices
## Call:cortest(R1 = educacion_prep)
## Chi Square value 1863.54 with df = 91 with probability < 00 < 0.05 (es correcto), entonces se rechaza Ho.
Esto quiere decir que la correlaciĆ³n entre variables es igual a 0
por ende no se crearĆan los componentes principales y esta data lista para la fase del modelado
Prueba de esfericidad de Barttlet
bartlett.test(educacion_prep)##
## Bartlett test of homogeneity of variances
##
## data: educacion_prep
## Bartlett's K-squared = 2645.8, df = 13, p-value < 2.2e-162.2e-16 < 0.05 (es correcto), entonces se rechaza Ho
Esto quiere decir que la matriz de correlaciones es distinta de la matriz de identidad
Prueba KMO (Kaiser Meyer Olkin)
KMO(educacion_prep)## Kaiser-Meyer-Olkin factor adequacy
## Call: KMO(r = educacion_prep)
## Overall MSA = 0.74
## MSA for each item =
## almain_1 almapr_2 almase_3 taasi_17 taasp_18 taass_19 taefp_22 taefs_23
## 0.79 0.88 0.77 0.54 0.51 0.80 0.49 0.43
## doin_24 dopr_25 dose_26 cein_29 cepr_30 cese_31
## 0.68 0.77 0.78 0.77 0.71 0.93Overall MSA = 0.74 > 0.5 (Se cumple),
Se justifica el anƔlisis de componentes principales
CASO 2: Distrito
Lectura de los datos
distritos<-read.csv("https://gist.githubusercontent.com/BenjiSantos/33fee8a00211146958990b66f864c70b/raw/a83679c2f26e016d7e71268badd388a5356d3db6/distritos.csv", sep = ";")
summary(distritos)## distrito ocu_vivi pobpjov sinelect
## Length:34 Min. :1.030 Min. :3.700 Min. : 0.350
## Class :character 1st Qu.:1.065 1st Qu.:4.525 1st Qu.: 1.775
## Mode :character Median :1.100 Median :5.100 Median : 4.355
## Mean :1.116 Mean :5.024 Mean : 9.687
## 3rd Qu.:1.150 3rd Qu.:5.475 3rd Qu.:16.310
## Max. :1.330 Max. :6.500 Max. :32.930
## sinagua pea1619 pocprin peam15
## Min. : 7.02 Min. :0.100 Min. : 1.100 Min. :15.63
## 1st Qu.:14.14 1st Qu.:0.725 1st Qu.: 4.125 1st Qu.:29.10
## Median :22.51 Median :1.700 Median : 6.650 Median :42.69
## Mean :25.17 Mean :2.162 Mean : 7.141 Mean :43.05
## 3rd Qu.:36.66 3rd Qu.:3.875 3rd Qu.:10.075 3rd Qu.:59.59
## Max. :65.76 Max. :5.400 Max. :14.400 Max. :65.11head(distritos)## distrito ocu_vivi pobpjov sinelect sinagua pea1619 pocprin peam15
## 1 Ate 1.15 5.3 27.60 51.10 3.9 1.1 63.48
## 2 Barranco 1.09 4.5 1.59 8.32 0.8 3.9 33.48
## 3 BreƱa 1.08 4.4 2.20 23.15 0.9 4.0 37.89
## 4 Carabayllo 1.10 5.1 30.13 38.09 4.5 12.6 63.65
## 5 Comas 1.20 5.9 10.92 24.27 3.8 9.4 60.37
## 6 Chorrillos 1.15 5.5 16.77 37.11 3.2 10.6 18.78Estandarisamos los datos para aplicar K-means,
para ellos es necesario quitar la variable cualitativa
distritosFilter=distritos[,2:8]
distritosStandar<-scale(distritosFilter)AplicanciĆ³n del mĆ©todo del codo para ver la cantidad Ć³ptima de clusters
fviz_nbclust(distritosStandar, kmeans, method = "wss")
AplicaciĆ³n del mĆ©todo silhouette para ver la cantidad Ć³ptima de clusters
fviz_nbclust(distritosStandar, FUNcluster=kmeans, method="silhouette")+theme_classic()