data <- read.csv(file = "~/Dropbox/DKE/Machine Learning/Master Assignment/Chronic_Kidney_Disease/chronic_kidney_disease_full.csv")
attach(data)
# Data Preprocessing
# Sort dataframe by numerical
# and categorical cols so writing
# imputation code will be easier
### Data preproc
data$al <- as.factor(data$al)
data$su <- as.factor(data$su)
nums <- sapply(data, is.numeric)
cat <- sapply(data, is.factor)
data_num <- data[,nums]
data_cat <- data[,cat]
rm(nums, cat)
data <- cbind(data_num, data_cat)
require(mice)
require(VIM)
require(lattice)
aggr_plot <- aggr(data[-c(25)],combination=T, sortVars=TRUE, labels=names(data), cex.axis=.9, gap=1, ylab=c("Histogram of missing data","Pattern"))
##
## Variables sorted by number of missings:
## Variable Count
## rbc 0.3800
## rbcc 0.3275
## wbcc 0.2650
## pot 0.2200
## sod 0.2175
## pcv 0.1775
## pc 0.1625
## hemo 0.1300
## su 0.1225
## sg 0.1175
## al 0.1150
## bgr 0.1100
## bu 0.0475
## sc 0.0425
## bp 0.0300
## age 0.0225
## pcc 0.0100
## ba 0.0100
## htn 0.0050
## dm 0.0050
## cad 0.0050
## appet 0.0025
## pe 0.0025
## ane 0.0025
# Number of rows without any missing value
nrow(data[complete.cases(data),])/nrow(data)
## [1] 0.395
marginplot(data[c("hemo","sc")])
Bottom Bar-plots are supposed to be quite similiar if data is MCAR.
##
## iter imp variable
## 1 1 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 2 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 3 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 4 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 5 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 6 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 7 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 8 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 9 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 10 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 11 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 12 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 13 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 14 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 15 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 16 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 17 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 18 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 19 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 1 20 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 1 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 2 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 3 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 4 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 5 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 6 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 7 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 8 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 9 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 10 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 11 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 12 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 13 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 14 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 15 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 16 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 17 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 18 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 19 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 2 20 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 1 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 2 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 3 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 4 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 5 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 6 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 7 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 8 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 9 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 10 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 11 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 12 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 13 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 14 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 15 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 16 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 17 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 18 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 19 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 3 20 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 1 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 2 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 3 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 4 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 5 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 6 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 7 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 8 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 9 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 10 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 11 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 12 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 13 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 14 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 15 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 16 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 17 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 18 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 19 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 4 20 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 1 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 2 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 3 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 4 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 5 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 6 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 7 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 8 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 9 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 10 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 11 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 12 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 13 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 14 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 15 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 16 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 17 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 18 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 19 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
## 5 20 age bp sg bgr bu sc sod pot hemo pcv wbcc rbcc al su rbc pc pcc ba htn dm cad appet pe ane
lattice::densityplot(data_imputed)