ds

setwd("D:\\TYCS\\B1\\DS_16\\DataSets")
getwd()
bike_data = read.csv("day.csv",header=TRUE)
boxplot(bike_data[,c('temp','atemp','hum','windspeed')])
median_value<-median(bike_data$hum)
high<-mean(bike_data$hum) + 2* sd(bike_data$hum)
low<-mean(bike_data$hum) - 2 * sd(bike_data$hum)
bike_data$hum<-ifelse(bike_data$hum>high | bike_data$hum<low, median_value, bike_data$hum)
boxplot(bike_data[,c('temp','atemp','hum','windspeed')])
median_value<-median(bike_data$windspeed)
sd(bike_data$windspeed)
high<-mean(bike_data$windspeed) + 2* sd(bike_data$windspeed)
low<-mean(bike_data$windspeed) - 2* sd(bike_data$windspeed)
bike_data$windspeed<-ifelse(bike_data$windspeed>high | bike_data$windspeed<low, median_value,bike_data$windspeed)

boxplot(bike_data[,c('temp','atemp','hum','windspeed')])

p-7

install.packages("caTools")

install.packages("dplyr")

install.packages("rpart")

install.packages("rpart.plot")

library(caTools)

library(dplyr)

library(rpart)

library(rpart.plot)

library(caret)

setwd("D:\\TYCS\\B1\\DS_16\\DataSets")

getwd()

x=read.csv("titanic.csv",header=TRUE)

newdata<-na.omit(x)

data=select(newdata,-Cabin)

sample_data=sample.split(data,SplitRatio=0.7)

train_data<-subset(data,sample_data==TRUE)

test_data<-subset(data,sample_data==FALSE)

Model<-glm(Survived~PassengerId+Pclass+Age+SibSp+Parch,data=train_data)

summary(Model)

predicted<-predict(Model,newdata=test_data)

test_data$predicted=ifelse(predicted>0.5,1,0)

actual=factor(test_data$Survived)

predicted=factor(test_data$predicted)

confusionMatrix(predicted,actual,mode='everything')

rtree <- rpart(Survived~PassengerId+Pclass+Age+SibSp+Parch,data=train_data)

y=predict(rtree,newdata=test_data)

test_data$y=ifelse(y>0.5,1,0)

rpart.plot(rtree,main="Decision Tree for Titanic Dataset")

a=factor(test_data$Survived)

p=factor(test_data$y)

confusionMatrix(p,a,mode='everything')

p-8

library(dplyr)

View(iris)

mydata=select(iris,c(1,2,3,4))

#cor(mydata)

mean(cor(mydata))

PCA=princomp(mydata)

summary(PCA)

PCA$loadings

PC=PCA$scores

cor(PC)

install.packages("factoextra")

library(factoextra)

get_eig(PCA)

fviz_eig(PCA,addlabels = TRUE)

fviz_pca_var(PCA,col.var="contrib")

fviz_pca_biplot(PCA,col.ind="Blue",geom="point")+labs(title="PCA",x="PC1",y="PC2")

p-6

#Linear regression

install.packages("caTools")

library(caTools)

head(mtcars)

a<- sample.split(mtcars$mpg,SplitRatio=0.7)

training_data <- mtcars[a,]

testing_data <- mtcars[!a,]

#testing_data

dim(training_data)

dim(testing_data)

plot(mpg~drat, data=mtcars)

Model=lm(mpg~drat, data=training_data)

t=summary(Model)

t

plot(mpg~drat, col="blue",cex=1.3,pch=16,data=training_data)

abline(Model)

#calculate MSE

mean(t$residuals^2)

Test=predict(Model,newdata=testing_data)

testing_data$Test=Test

View(testing_data)

result=predict(Model,data.frame(drat=3.90))

print(result)

#multiple regression

library(caret)

data(mtcars)

head(mtcars)

in_train <- createDataPartition(y=mtcars$mpg,p=0.7,list=FALSE)

training_data <- mtcars[in_train,]

testing_data <- mtcars[-in_train,]

Model=lm(mpg~cyl+disp,data=trainig_data)

model_summ<-summary(Model)

model_summ

#calculate MSE

mean(model_summ$residuals^2)

Test=predict(Model,newdata=testing_data)

testing_data$Test=Test

View(testing_data)

new<-data.frame(cyl=c(6),disp=c(160))

results=predict(Model,newdata=new)

print(results)