Bivariado

0.1 Correlaciones

Tomamos la base de datos del artículo:

library(rio)
recall<-import("https://github.com/jincio/Imperfect_recall/raw/master/recalldataset.dta")

names(recall)

##  [1] "idmun"          "region"         "voters"         "request"       
##  [5] "vote"           "removal"        "enp"            "competition"   
##  [9] "indigenous"     "execution"      "budget"         "canon"         
## [13] "previous"       "conflict"       "incumbent"      "share"         
## [17] "rural"          "period"         "days"           "firstexecution"
## [21] "female"         "poverty2"       "investment"

Existirá correlación entre el la competitividad y las regalias mineras

cor(recall$canon,recall$execution,use = "complete.obs")

## [1] -0.2587664

cor(recall$canon,recall$execution,use = "complete.obs", method = "kendall")

## [1] -0.2253771

cor.test(recall$canon,recall$execution)

## 
##  Pearson's product-moment correlation
## 
## data:  recall$canon and recall$execution
## t = -18.739, df = 4893, p-value < 2.2e-16
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.2847175 -0.2324364
## sample estimates:
##        cor 
## -0.2587664

cor(recall[,7:12], use= "complete.obs")

##                     enp competition  indigenous   execution      budget
## enp          1.00000000 -0.41369220  0.11892046 -0.07563594 -0.10379910
## competition -0.41369220  1.00000000 -0.06243175  0.02768176  0.03305790
## indigenous   0.11892046 -0.06243175  1.00000000 -0.06564584  0.08987345
## execution   -0.07563594  0.02768176 -0.06564584  1.00000000 -0.29726150
## budget      -0.10379910  0.03305790  0.08987345 -0.29726150  1.00000000
## canon       -0.05402253  0.01635722  0.07501678 -0.25869880  0.95052677
##                   canon
## enp         -0.05402253
## competition  0.01635722
## indigenous   0.07501678
## execution   -0.25869880
## budget       0.95052677
## canon        1.00000000

library(Hmisc)

## Loading required package: lattice

## Loading required package: survival

## Loading required package: Formula

## Loading required package: ggplot2

## 
## Attaching package: 'Hmisc'

## The following objects are masked from 'package:base':
## 
##     format.pval, units

rcorr(as.matrix(recall[,7:12]))

##               enp competition indigenous execution budget canon
## enp          1.00       -0.42       0.13     -0.08  -0.15 -0.05
## competition -0.42        1.00      -0.07      0.03   0.04  0.02
## indigenous   0.13       -0.07       1.00     -0.07   0.08  0.08
## execution   -0.08        0.03      -0.07      1.00  -0.30 -0.26
## budget      -0.15        0.04       0.08     -0.30   1.00  0.95
## canon       -0.05        0.02       0.08     -0.26   0.95  1.00
## 
## n
##              enp competition indigenous execution budget canon
## enp         6558        6555       6533      4895   6518  4917
## competition 6555        6555       6530      4892   6515  4914
## indigenous  6533        6530       6533      4878   6498  4900
## execution   4895        4892       4878      4895   4895  4895
## budget      6518        6515       6498      4895   6518  4895
## canon       4917        4914       4900      4895   4895  4917
## 
## P
##             enp    competition indigenous execution budget canon 
## enp                0.0000      0.0000     0.0000    0.0000 0.0001
## competition 0.0000             0.0000     0.0448    0.0015 0.2560
## indigenous  0.0000 0.0000                 0.0000    0.0000 0.0000
## execution   0.0000 0.0448      0.0000               0.0000 0.0000
## budget      0.0000 0.0015      0.0000     0.0000           0.0000
## canon       0.0001 0.2560      0.0000     0.0000    0.0000

0.1.1 Correlaciones por grupo usando dplyr

library(dplyr)

## 
## Attaching package: 'dplyr'

## The following objects are masked from 'package:Hmisc':
## 
##     src, summarize

## The following objects are masked from 'package:stats':
## 
##     filter, lag

## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

recall %>% group_by(period) %>%
              do(as.data.frame(cor(.[,c("canon","budget")], method="spearman", use="pairwise.complete.obs")))

## # A tibble: 8 x 3
## # Groups:   period [4]
##   period  canon budget
##    <dbl>  <dbl>  <dbl>
## 1      1 NA     NA    
## 2      1 NA      1    
## 3      2  1      0.855
## 4      2  0.855  1    
## 5      3  1      0.767
## 6      3  0.767  1.000
## 7      4  1      0.666
## 8      4  0.666  1

0.2 Pruebas T

Pruebas de diferencias de medias:

t.test(recall$enp~recall$incumbent)

## 
##  Welch Two Sample t-test
## 
## data:  recall$enp by recall$incumbent
## t = 18.428, df = 2060.1, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.7307678 0.9048306
## sample estimates:
## mean in group 0 mean in group 1 
##        4.682983        3.865184

t.test(recall$budget,recall$canon)

## 
##  Welch Two Sample t-test
## 
## data:  recall$budget and recall$canon
## t = 12.658, df = 10563, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##   758.8097 1036.8967
## sample estimates:
## mean of x mean of y 
##  2218.847  1320.994

t.test(recall$enp, mu=4.5)

## 
##  One Sample t-test
## 
## data:  recall$enp
## t = 1.6325, df = 6557, p-value = 0.1026
## alternative hypothesis: true mean is not equal to 4.5
## 95 percent confidence interval:
##  4.493581 4.570356
## sample estimates:
## mean of x 
##  4.531969

0.3 Chi-square

chisq.test(table(recall$vote, recall$incumbent))

## 
##  Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  table(recall$vote, recall$incumbent)
## X-squared = 2.7529, df = 1, p-value = 0.09708

0.4 Gráficos Bivariados

0.4.1 Correlaciones

#install.packages("corrplot")
library(corrplot)

## corrplot 0.84 loaded

corrplot(cor(recall[,7:12], use= "complete.obs"), type = "upper", order = "hclust", 
         tl.col = "black", tl.srt = 45)

correlations<-rcorr(as.matrix(recall[,7:12]))

corrplot(correlations$r, type="upper", order="hclust", 
         p.mat = correlations$P, sig.level = 0.001, insig = "blank")

col<- colorRampPalette(c("blue", "white", "red"))(20)
col

##  [1] "#0000FF" "#1A1AFF" "#3535FF" "#5050FF" "#6B6BFF" "#8686FF" "#A1A1FF"
##  [8] "#BBBBFF" "#D6D6FF" "#F1F1FF" "#FFF1F1" "#FFD6D6" "#FFBBBB" "#FFA1A1"
## [15] "#FF8686" "#FF6B6B" "#FF5050" "#FF3535" "#FF1A1A" "#FF0000"

col<- colorRampPalette(c("blue", "white", "red"))(6)
heatmap(cor(recall[,7:12], use= "complete.obs"), col = col, symm = TRUE)

0.4.2 Correlations plots with ggplot

correlations<-round(cor(recall[,7:12], use= "complete.obs"),3)
correlations

##                enp competition indigenous execution budget  canon
## enp          1.000      -0.414      0.119    -0.076 -0.104 -0.054
## competition -0.414       1.000     -0.062     0.028  0.033  0.016
## indigenous   0.119      -0.062      1.000    -0.066  0.090  0.075
## execution   -0.076       0.028     -0.066     1.000 -0.297 -0.259
## budget      -0.104       0.033      0.090    -0.297  1.000  0.951
## canon       -0.054       0.016      0.075    -0.259  0.951  1.000

Necesitamos la data en otro formato:

reshape2::melt(correlations)

##           Var1        Var2  value
## 1          enp         enp  1.000
## 2  competition         enp -0.414
## 3   indigenous         enp  0.119
## 4    execution         enp -0.076
## 5       budget         enp -0.104
## 6        canon         enp -0.054
## 7          enp competition -0.414
## 8  competition competition  1.000
## 9   indigenous competition -0.062
## 10   execution competition  0.028
## 11      budget competition  0.033
## 12       canon competition  0.016
## 13         enp  indigenous  0.119
## 14 competition  indigenous -0.062
## 15  indigenous  indigenous  1.000
## 16   execution  indigenous -0.066
## 17      budget  indigenous  0.090
## 18       canon  indigenous  0.075
## 19         enp   execution -0.076
## 20 competition   execution  0.028
## 21  indigenous   execution -0.066
## 22   execution   execution  1.000
## 23      budget   execution -0.297
## 24       canon   execution -0.259
## 25         enp      budget -0.104
## 26 competition      budget  0.033
## 27  indigenous      budget  0.090
## 28   execution      budget -0.297
## 29      budget      budget  1.000
## 30       canon      budget  0.951
## 31         enp       canon -0.054
## 32 competition       canon  0.016
## 33  indigenous       canon  0.075
## 34   execution       canon -0.259
## 35      budget       canon  0.951
## 36       canon       canon  1.000

Se puede llamar la función de un paquete usando ::. Es útil en dos momentos: si no vas a usar el paquete nuevamente o si hay conflicto con el nombre de las funciones.

reshape2::melt(correlations)%>%ggplot(aes(x=Var1, y=Var2, fill=value)) + geom_tile()

correlations[lower.tri(correlations)] <- NA
correlations

##             enp competition indigenous execution budget  canon
## enp           1      -0.414      0.119    -0.076 -0.104 -0.054
## competition  NA       1.000     -0.062     0.028  0.033  0.016
## indigenous   NA          NA      1.000    -0.066  0.090  0.075
## execution    NA          NA         NA     1.000 -0.297 -0.259
## budget       NA          NA         NA        NA  1.000  0.951
## canon        NA          NA         NA        NA     NA  1.000

melt(correlations, na.rm = TRUE)

##           Var1        Var2  value
## 1          enp         enp  1.000
## 7          enp competition -0.414
## 8  competition competition  1.000
## 13         enp  indigenous  0.119
## 14 competition  indigenous -0.062
## 15  indigenous  indigenous  1.000
## 19         enp   execution -0.076
## 20 competition   execution  0.028
## 21  indigenous   execution -0.066
## 22   execution   execution  1.000
## 25         enp      budget -0.104
## 26 competition      budget  0.033
## 27  indigenous      budget  0.090
## 28   execution      budget -0.297
## 29      budget      budget  1.000
## 31         enp       canon -0.054
## 32 competition       canon  0.016
## 33  indigenous       canon  0.075
## 34   execution       canon -0.259
## 35      budget       canon  0.951
## 36       canon       canon  1.000

melt(correlations, na.rm = TRUE)%>%
  ggplot(aes(Var2, Var1, fill = value))+
 geom_tile(color = "white")+
 scale_fill_gradient2(low = "blue", high = "red", mid = "white", 
   midpoint = 0, limit = c(-1,1), space = "Lab", 
   name="Pearson\nCorrelation")

melt(correlations, na.rm = TRUE)%>%
  ggplot(aes(Var2, Var1, fill = value))+
 geom_tile(color = "white")+
 scale_fill_gradient2(low = "blue", high = "red", mid = "white", 
   midpoint = 0, limit = c(-1,1), space = "Lab", 
   name="Pearson\nCorrelation") +
  theme_minimal()#+

 # theme(axis.text.x = element_text(angle = 45, vjust = 1, 
  #  size = 12, hjust = 1))+ ### Pone los nombres de las variables inclinadas
 #coord_fixed()

melt(correlations, na.rm = TRUE)%>%
  ggplot(aes(Var2, Var1, fill = value))+
 geom_tile(color = "white")+
 scale_fill_gradient2(low = "blue", high = "red", mid = "white", 
   midpoint = 0, limit = c(-1,1), space = "Lab", 
   name="Pearson\nCorrelation") +
  theme_minimal()+
  theme(axis.text.x = element_text(angle = 45, vjust = 1, 
    size = 12, hjust = 1))

0.4.3 Error bars

recall%>%select(incumbent,enp)%>%group_by(incumbent)%>%
  summarise(mean=mean(enp), sd=sd(enp))%>%
  ggplot(aes(x=incumbent, y=mean, group=incumbent))+
  geom_line()+
  geom_point()+
  geom_errorbar(aes(ymin=mean-sd, ymax=mean+sd), width=.2,
                 position=position_dodge(0.05))

## geom_path: Each group consists of only one observation. Do you need to
## adjust the group aesthetic?

library(gplots)
plotmeans(enp ~ incumbent, data = recall, mean.labels = TRUE)