IST687_Final_Project.Rmd

---
title: "IST 687 Final Project"
author: "Omkar Sabnis"
date: "2022-12-03"
output:
  word_document: default
  html_document: default
---
```{r}
#install.packages('rsconnect')
library(rsconnect)
```

```{r}
#Librarying all the required packages
library(dplyr)
library(ggplot2)
library(tidyverse)
#install.packages("rsample")
#library(rsample)
#install.packages("caret")
#install.packages("kernlab")
#install.packages("e1071")
#install.packages("arules")
#install.packages("arulesViz")
#install.packages("imputeTS")
#install.packages("rio")
#install.packages("rpart")
#install.packages("rpart.plot")
library(caret)
library(kernlab)
library(e1071)
library(arules)
library(arulesViz)
library(imputeTS)
library(rio)
library(rpart)
library(rpart.plot)
#install.packages("shiny")
#install.packages("shinydashboard")
library(shiny)
#library(shinydashboard)
```


```{r}
library(tidyverse)
# Loading the dataset
data1 <- read_csv("https://intro-datascience.s3.us-east-2.amazonaws.com/HMO_data.csv")
data <- data1
head(data)
```


```{r}
# Exploring the data
view(data)
str(data)
summary(data)
```



```{r}
# Performing Data Cleaning
# Dealing with N/A values 
colSums(is.na(data))
anyNA(data)
data$bmi<- na_interpolation(data$bmi)
data$hypertension <- na_interpolation(data$hypertension)

boxplot(data$cost)
quartiles <- quantile(data$cost, probs=c(.25, .75), na.rm = FALSE)
IQR <- IQR(data$cost)
Lower <- quartiles[1] - 1.5*IQR
Upper <- quartiles[2] + 1.5*IQR 
data <- subset(data, data$cost > Lower & data$cost < Upper)
data$expensive <- data$cost > 5000

```


```{r}
#Defining a variable "expensive" which has costs more than 4000 
quantile(data$cost, probs = c(0.75))
data$exp <- data$cost>4000
# replacing TRUE with 1 and FALSE with 0
data <- data %>% mutate(exp = str_replace_all( string = exp, pattern = "TRUE", "1"))
data <- data %>% mutate(exp = str_replace_all( string = exp, pattern = "FALSE", "0"))
data
```

```{r}
#Adding the variables that are important for analysis in a new dataframe "HMO"
HMO<- data.frame(age = data$age,
                      bmi = data$bmi,
                      smoker= data$smoker,
                      yearly_physical= data$yearly_physical,
                      exercise =data$exercise,
                      hypertension = data$hypertension,
                      exp=as.factor(data$exp))
  
#viewing the new data frame
str(HMO)
```

```{r}
# Dividing data into training set and testing sets
trainL <- createDataPartition(y=HMO$exp,p=0.85,list=FALSE)
trainS <- HMO[trainL,]
testS <- HMO[-trainL,]
dim(trainS)
dim(testS)
library(readr)
library(data.table)
write.csv(trainS, "C:\\Users\\omkar\\Desktop\\IST 687 Introduction to Data Science\\Final Project\\DATA To IMPORT\\trainset.csv", row.names=FALSE)
write.csv(testS, "C:\\Users\\omkar\\Desktop\\IST 687 Introduction to Data Science\\Final Project\\DATA To IMPORT\\InputFile.csv", row.names=FALSE)
```

```{r}
# Creating SVM model
ksvm_model <- ksvm(data= trainS, exp~.,C=5, CV=3, prob.model= TRUE)
SVMPred<- predict(ksvm_model,newdata= testS, type= "response")
head(SVMPred)

# Checking accuracy of ksvm model using confusion matrix
confusionMatrix(SVMPred,testS$exp)
```

```{r}
# Building a tree model
rpart_model <- rpart(exp ~., data = trainS, method = "class")
rpartPred <- predict(rpart_model, newdata= testS, type= "class")
head(rpartPred)
confusionMatrix(rpartPred, testS$exp)
```

```{r}
# Linear model
trainS$exp<-as.numeric(trainS$exp)
testS$exp<-as.numeric(testS$exp)
lmOut <- lm(exp~.,data=trainS)
summary(lmOut)
#predict(lmOut,testS,type = 'response')
```




```{r}
#dividing expensive and inexpensive people into 2 subsets
expPeople <- subset(data,exp=="1")
inexpPeople <- subset(data,exp=="0") 
head(expPeople)
head(inexpPeople)
```

```{r}
#Visualizations
hist(expPeople$age)
hist(inexpPeople$age)
hist(data$bmi)
hist(data$hypertension)
hist(data$cost)
```
```{r}
#Visualizations
#Location by max cost in each state
#Maps(Cost based on location)
dfAgg <- data %>% group_by(location) %>% summarise(total_cost = max(cost))
dfAgg$state <- tolower(dfAgg$location)
us <- map_data("state")
us$state <- us$region
mergedNew <- merge(dfAgg,us,on = "state")
mergedNew <- mergedNew[order(mergedNew$order),]
map <- ggplot(mergedNew) + geom_polygon(aes(x = long, y = lat, group = group,fill = total_cost), color = "black") 
map + scale_fill_continuous(low = "black", high = "purple", name = "total_cost", label = scales::comma) + coord_map() +ggtitle(" Mappping the maximum cost per state for the expensive and non expensive people")
```
```{r}
#Visualizations
#Location by Average age in each state
dfAgg <- data %>% group_by(location) %>% summarise(avg_age= mean(age))
dfAgg$state <- tolower(dfAgg$location)
us <- map_data("state")
us$state <- us$region
mergedNew <- merge(dfAgg,us,on = "state")
mergedNew <- mergedNew[order(mergedNew$order),]
map <- ggplot(mergedNew) + geom_polygon(aes(x = long, y = lat, group = group,fill = avg_age), color = "black") 
map + scale_fill_continuous(low = "white", high = "red", name = "avg_age", label = scales::comma) + coord_map() +ggtitle(" Mappping the maximum cost per state for the expensive and non expensive people")
```
```{r}
#Visualizations

boxplot(data$bmi)
boxplot(data$age)
boxplot(data$hypertension)
boxplot(data$cost)
```
```{r}
#Visualizations
#scatter Plot age vs cost
bmicost<- ggplot(data, aes(x=bmi,y=cost,color=location_type)) +geom_point()
bmicost
agecost<- ggplot(data, aes(x=age,y=cost,group=exercise)) +geom_point(aes(color=exercise)) + scale_color_manual(values=c('purple','black'))
agecost

```


```{r}
hist(expPeople$cost)
hist(inexpPeople$cost)
```

```{r}
library(shiny)
library(ggplot2)
library(plotly)
library(rmarkdown)
library(knitr)
library(pander)
```