User Knowledge Modeling Data Set.Rmd

---
title: "User Knowledge Modeling Data Set"
author: "2015015017 park ji su"
date: "2017년 11월 13일"
output: html_document
---

```{r setup, include=FALSE}

#knn

user <- read.csv("user.csv", stringsAsFactors = FALSE)

str(user)

table(user$UNS)

round(prop.table(table(user$UNS)) * 100, digits = 1)

normalize <- function(x) {
 return ((x - min(x)) / (max(x) - min(x)))
}

user_n<-as.data.frame(lapply(user[1:5],normalize))

user_train <- user_n[1:198, ]
user_test <- user_n[199:258, ]


user_train_labels <- user[1:198, 6]
user_test_labels <- user[199:258, 6]

library(class)

user_test_pred <- knn(train = user_train, test = user_test,
                      cl = user_train_labels, k=8)

# 7  is bes 83%
library(gmodels)


CrossTable(x = user_test_labels, y = user_test_pred,
           prop.chisq=FALSE)


#desicion tree


user <- read.csv("user.csv", stringsAsFactors = FALSE)

str(user)

set.seed(123)

train_sample <-sample(258,198)
str(train_sample)

user_train <- user[train_sample,-6]
user_test <- user[-train_sample, ]

user_train_label<- user[train_sample, 6]
user_train_label<- as.factor(user_train_label)

#install.packages("C50")
library(C50)

user_model <- C5.0(user_train,user_train_label)

summary(user_model)

user_pred<-predict(user_model,user_test)


CrossTable(user_test$UNS,user_pred,prop.chisq=FALSE,prop.c = FALSE,prop.r = FALSE,dnn= c('actual default','predicted default'))

# 93%
```