install.packages("class")

install.packages("gmodels")

install.packages("vegan")

install.packages("e1071")

library(class)

library(gmodels)

setwd('/Users/yiqian/Documents/GitHub/BigDataProject2')

mr<-read.table("agaricus-lepiota.data", sep=",", header=FALSE)

str(mrdata)

mr.df<-as.data.frame(mrdata)

mr.df

names(mr)<-c('class', 'cshape', 'csurface', 'ccolor', 'bruises', 'odor', 'gattach', 'gspace', 'gsize',

'gcolor', 'sshape', 'sroot', 'ssabove', 'ssbelow', 'scabove', 'scbelow', 'vtype', 'vcolor',

'rnumber', 'rtype', 'spcolor', 'popnum', 'habitat')

names(mr)

mrf<-as.data.frame(mr,stringAsFactors=FALSE)

mrf[,c(1,23)]<-sapply(mrf[,c(1,23)], as.character)

# class

mrf$class[mrf$class=='e']<-0

mrf$class[mrf$class=='p']<-1

# cap-shape

mrf$cshape<-as.character(mrf$cshape)

mrf$cshape[mrf$cshape=='b']<-1

mrf$cshape[mrf$cshape=='c']<-3

mrf$cshape[mrf$cshape=='x']<-5

mrf$cshape[mrf$cshape=='f']<-7

mrf$cshape[mrf$cshape=='k']<-9

mrf$cshape[mrf$cshape=='s']<-11

# cap-surface

mrf$csurface<-as.character(mrf$csurface)

mrf$csurface[mrf$csurface=='f']<-1

mrf$csurface[mrf$csurface=='g']<-3

mrf$csurface[mrf$csurface=='y']<-5

mrf$csurface[mrf$csurface=='s']<-7

# cap-color

mrf$ccolor<-as.character(mrf$ccolor)

mrf$ccolor[mrf$ccolor=='n']<-1

mrf$ccolor[mrf$ccolor=='b']<-3

mrf$ccolor[mrf$ccolor=='c']<-5

mrf$ccolor[mrf$ccolor=='g']<-7

mrf$ccolor[mrf$ccolor=='r']<-9

mrf$ccolor[mrf$ccolor=='p']<-11

mrf$ccolor[mrf$ccolor=='u']<-13

mrf$ccolor[mrf$ccolor=='e']<-15

mrf$ccolor[mrf$ccolor=='w']<-17

mrf$ccolor[mrf$ccolor=='y']<-19

# bruises

mrf$bruises<-as.character(mrf$bruises)

mrf$bruises[mrf$bruises=='t']<-1

mrf$bruises[mrf$bruises=='f']<-3

# odor

mrf$odor<-as.character(mrf$odor)

mrf$odor[mrf$odor=='a']<-1

mrf$odor[mrf$odor=='l']<-3

mrf$odor[mrf$odor=='c']<-5

mrf$odor[mrf$odor=='y']<-7

mrf$odor[mrf$odor=='f']<-9

mrf$odor[mrf$odor=='m']<-11

mrf$odor[mrf$odor=='n']<-13

mrf$odor[mrf$odor=='p']<-15

mrf$odor[mrf$odor=='s']<-17

# gattach

mrf$gattach<-as.character(mrf$gattach)

mrf$gattach[mrf$gattach=='a']<-1

mrf$gattach[mrf$gattach=='d']<-3

mrf$gattach[mrf$gattach=='f']<-5

mrf$gattach[mrf$gattach=='n']<-7

# gspace

mrf$gspace<-as.character(mrf$gspace)

mrf$gspace[mrf$gspace=='c']<-1

mrf$gspace[mrf$gspace=='w']<-3

mrf$gspace[mrf$gspace=='d']<-5

# gsize

mrf$gsize<-as.character(mrf$gsize)

mrf$gsize[mrf$gsize=='b']<-1

mrf$gsize[mrf$gsize=='n']<-3

# gcolor

mrf$gcolor<-as.character(mrf$gcolor)

mrf$gcolor[mrf$gcolor=='k']<-1

mrf$gcolor[mrf$gcolor=='n']<-3

mrf$gcolor[mrf$gcolor=='b']<-5

mrf$gcolor[mrf$gcolor=='h']<-7

mrf$gcolor[mrf$gcolor=='g']<-9

mrf$gcolor[mrf$gcolor=='r']<-11

mrf$gcolor[mrf$gcolor=='o']<-13

mrf$gcolor[mrf$gcolor=='p']<-15

mrf$gcolor[mrf$gcolor=='u']<-17

mrf$gcolor[mrf$gcolor=='e']<-19

mrf$gcolor[mrf$gcolor=='w']<-21

mrf$gcolor[mrf$gcolor=='y']<-23

# sshape

mrf$sshape<-as.character(mrf$sshape)

mrf$sshape[mrf$sshape=='e']<-1

mrf$sshape[mrf$sshape=='t']<-3

# sroot

mrf$sroot<-as.character(mrf$sroot)

mrf$sroot[mrf$sroot=='b']<-1

mrf$sroot[mrf$sroot=='c']<-3

mrf$sroot[mrf$sroot=='u']<-5

mrf$sroot[mrf$sroot=='e']<-7

mrf$sroot[mrf$sroot=='z']<-9

mrf$sroot[mrf$sroot=='r']<-11

mrf$sroot[mrf$sroot=='?']<-13

# ssabove

mrf$ssabove<-as.character(mrf$ssabove)

mrf$ssabove[mrf$ssabove=='f']<-1

mrf$ssabove[mrf$ssabove=='y']<-3

mrf$ssabove[mrf$ssabove=='k']<-5

mrf$ssabove[mrf$ssabove=='s']<-7

# ssbelow

mrf$ssbelow<-as.character(mrf$ssbelow)

mrf$ssbelow[mrf$ssbelow=='f']<-1

mrf$ssbelow[mrf$ssbelow=='y']<-3

mrf$ssbelow[mrf$ssbelow=='k']<-5

mrf$ssbelow[mrf$ssbelow=='s']<-7

# scabove

mrf$scabove<-as.character(mrf$scabove)

mrf$scabove[mrf$scabove=='n']<-1

mrf$scabove[mrf$scabove=='b']<-3

mrf$scabove[mrf$scabove=='c']<-5

mrf$scabove[mrf$scabove=='g']<-7

mrf$scabove[mrf$scabove=='o']<-9

mrf$scabove[mrf$scabove=='p']<-11

mrf$scabove[mrf$scabove=='e']<-13

mrf$scabove[mrf$scabove=='w']<-15

mrf$scabove[mrf$scabove=='y']<-17

# scbelow

mrf$scbelow<-as.character(mrf$scbelow)

mrf$scbelow[mrf$scbelow=='n']<-1

mrf$scbelow[mrf$scbelow=='b']<-3

mrf$scbelow[mrf$scbelow=='c']<-5

mrf$scbelow[mrf$scbelow=='g']<-7

mrf$scbelow[mrf$scbelow=='o']<-9

mrf$scbelow[mrf$scbelow=='p']<-11

mrf$scbelow[mrf$scbelow=='e']<-13

mrf$scbelow[mrf$scbelow=='w']<-15

mrf$scbelow[mrf$scbelow=='y']<-17

# vtype

mrf$vtype<-as.character(mrf$vtype)

mrf$vtype[mrf$vtype=='p']<-1

mrf$vtype[mrf$vtype=='u']<-3

# vcolor

mrf$vcolor<-as.character(mrf$vcolor)

mrf$vcolor[mrf$vcolor=='n']<-1

mrf$vcolor[mrf$vcolor=='o']<-3

mrf$vcolor[mrf$vcolor=='w']<-5

mrf$vcolor[mrf$vcolor=='y']<-7

# rnumber

mrf$rnumber<-as.character(mrf$rnumber)

mrf$rnumber[mrf$rnumber=='n']<-1

mrf$rnumber[mrf$rnumber=='o']<-3

mrf$rnumber[mrf$rnumber=='t']<-5

# rtype

mrf$rtype<-as.character(mrf$rtype)

mrf$rtype[mrf$rtype=='c']<-1

mrf$rtype[mrf$rtype=='e']<-3

mrf$rtype[mrf$rtype=='f']<-5

mrf$rtype[mrf$rtype=='l']<-7

mrf$rtype[mrf$rtype=='n']<-9

mrf$rtype[mrf$rtype=='p']<-11

mrf$rtype[mrf$rtype=='s']<-13

mrf$rtype[mrf$rtype=='z']<-15

# spcolor

mrf$spcolor<-as.character(mrf$spcolor)

mrf$spcolor[mrf$spcolor=='k']<-1

mrf$spcolor[mrf$spcolor=='n']<-3

mrf$spcolor[mrf$spcolor=='b']<-5

mrf$spcolor[mrf$spcolor=='h']<-7

mrf$spcolor[mrf$spcolor=='r']<-9

mrf$spcolor[mrf$spcolor=='o']<-11

mrf$spcolor[mrf$spcolor=='u']<-13

mrf$spcolor[mrf$spcolor=='w']<-15

mrf$spcolor[mrf$spcolor=='y']<-17

# popnum

mrf$popnum<-as.character(mrf$popnum)

mrf$popnum[mrf$popnum=='a']<-1

mrf$popnum[mrf$popnum=='c']<-3

mrf$popnum[mrf$popnum=='n']<-5

mrf$popnum[mrf$popnum=='s']<-7

mrf$popnum[mrf$popnum=='v']<-9

mrf$popnum[mrf$popnum=='y']<-11

# habitat

mrf$habitat<-as.character(mrf$habitat)

mrf$habitat[mrf$habitat=='g']<-1

mrf$habitat[mrf$habitat=='l']<-3

mrf$habitat[mrf$habitat=='m']<-5

mrf$habitat[mrf$habitat=='p']<-7

mrf$habitat[mrf$habitat=='u']<-9

mrf$habitat[mrf$habitat=='w']<-11

mrf$habitat[mrf$habitat=='d']<-13

dataset.df<-sample(nrow(mrf), 1*nrow(mrf))

## training-test dataset

train.df<-sample(nrow(mrf), 0.7*nrow(mrf))

#train.df<-sample(nrow(mrf), 0.6*nrow(mrf))

#train.df<-sample(nrow(mrf), 0.5*nrow(mrf))

#train set

mrftrain.df<-mrf[train.df,]

#mrftrain.df<-mrf[train.df,]

#mrftrain.df<-mrf[train.df,]

#mrftest set

mrftest.df<-mrf[-train.df,]

#mrftest.df<-mrf[-train.df,]

#mrftest.df<-mrf[-train.df,]

# k-means (no need train and test dataset)

kc3<-kmeans(dataset.df,3)

#kc5<-kmeans(dataset.df,5)

#kc7<-kmeans(dataset.df,7)

# knn

# get train label

knn_train_label<-mrftrain.df[,1]

# remove label

knn_train_data<-mrftrain.df[,2:23]

knn_test_data<-mrftest.df[,2:23]

# return prediction of test set

knn_pred<-knn(train=knn_train_data, test=knn_test_data, cl=knn_train_label, 24)

# eveluating knn

knn_test_label<-mrftest.df[,1]

CrossTable(x=knn_test_label, y=knn_pred, prop.chisq=FALSE)

# Hierarchical clustering

library(vegan)

distance.ex<-vegdist(mrftrain.df, method="euc", na.rm=TRUE)

hclust.ex<-hclust(distance.ex, method="ward.D2")

plot(hclust.ex,hang=-1)

pairs(class~cshape+csurface+ccolor+bruises+odor, data=mr)

# lm model

lm_model<-lm(mrftrain.df$ccolor ~ mrftrain.df$gspace, data=mrftrain.df)

# lm predict

lm_pred <- predict(lm_model, data=mrftest.df)

#plot(class~sshape+sroot, data=mr)

plot(mr$ccolor, mr$gspace, xlab = "ccolor", ylab='gspace',

main="ccolor-gspace graph")

#glm

glm_res<-glm(as.numeric(mrf$ccolor) ~ as.numeric(mrf$gspace) + as.numeric(mrf$gsize) +

as.numeric(mrf$gattach) + as.numeric(mrf$bruises), family = gaussian, data = mrf)

summary(glm_res)

# SVM separate uncertain dataset

library(e1071)

# get uncertain dataset

uncertain <- mrf[mrf$class==1,]

uncertain.df<-sample(nrow(uncertain), 1*nrow(uncertain))

svm_res <- svm(as.numeric(uncertain$ccolor) ~ as.numeric(uncertain$gspace) +

as.numeric(uncertain$gsize) + as.numeric(uncertain$gattach) +

as.numeric(uncertain$bruises), data=uncertain, kernel="linear")