else return(0)
}
#paritas(v)
d = apply(v, 1, paritas)
data = data.frame(v, factor(d), 3)
print(data)
data = data.frame(v, factor(d))
# parits problma
v = bincombinations(6)
paritas = function(v){
par = 0
for(x in v){
if(x %% 2 == 1){par = par+1}
}
if(par %% 2 == 1)return(1)
else return(0)
}
#paritas(v)
d = apply(v, 1, paritas)
data = data.frame(v, factor(d))
print(data)
# parits problma
v = bincombinations(10)
paritas = function(v){
par = 0
for(x in v){
if(x %% 2 == 1){par = par+1}
}
if(par %% 2 == 1)return(1)
else return(0)
}
#paritas(v)
d = apply(v, 1, paritas)
data = data.frame(v, factor(d))
print(data)
# parits problma
v = bincombinations(15)
paritas = function(v){
par = 0
for(x in v){
if(x %% 2 == 1){par = par+1}
}
if(par %% 2 == 1)return(1)
else return(0)
}
#paritas(v)
d = apply(v, 1, paritas)
data = data.frame(v, factor(d))
print(data)
# parits problma
v = bincombinations(5)
paritas = function(v){
par = 0
for(x in v){
if(x %% 2 == 1){par = par+1}
}
if(par %% 2 == 1)return(1)
else return(0)
}
#paritas(v)
d = apply(v, 1, paritas)
data = data.frame(v, factor(d))
print(data)
# parits problma
v = bincombinations(5)
paritas = function(v){
par = 0
for(x in v)
par = par+x
if(par %% 2 == 1)return(1)
else return(0)
}
#paritas(v)
d = apply(v, 1, paritas)
data = data.frame(v, factor(d))
print(data)
# parits problma
v = bincombinations(5)
paritas = function(v){
par = 0
for(x in v)
par = par + x
if(par %% 2 == 1)
return(1)
else 
return(0)
}
#paritas(v)
d = apply(v, 1, paritas)
data = data.frame(v, factor(d))
print(data)
# parits problma
v = bincombinations(3)
paritas = function(v){
par = 0
for(x in v)
par = par + x
if(par %% 2 == 1)
return(1)
else 
return(0)
}
#paritas(v)
d = apply(v, 1, paritas)
data = data.frame(v, factor(d))
print(data)
ceiling((5 + 1)/2)
3 %% 2
5 %% 2
q()
library(MASS)
A = mvrnorm(n=500, c(0,0), diag(2))
B = mvrnorm(n=500, c(2,0), matrix(c(4,0,0,4), 2))
A = cbind(A, rep(1, 500))
B = cbind(B, rep(1, 500))
data = data.frame(rbind(A,B))
library(nnet)
net = nnet(d~x1+x2, data=data, size=2, maxit=200)
t = table(
y=predict(net, type="class")
d=data$d
)
t
library(MASS)
A = mvrnorm(n=500, c(0,0), diag(2))
B = mvrnorm(n=500, c(2,0), matrix(c(4,0,0,4), 2))
C = mvrnorm(n=500, c(1,4), matrix(c(2,0,0,2), 2))
A = cbind(A, rep(1, 500))
B = cbind(B, rep(1, 500))
C = cbind(C, rep(1, 500))
data = data.frame(rbind(A,B,C))
names(data) = c("x1", "x2", "d")
library(nnet)
net = nnet(d~x1+x2, data=data, size=2, maxit=200)
t = table(
y=predict(net, type="class")
d=data$d
)
t
getwd()
setwd("D:\Zoli2\Prog\R")
setwd("D:/Zoli2/Prog/R")
getwd()
# osztlyozs (nem szerinti)
#setwd("D:/Zoli2/Prog/R")
data = read.csv("crabdata.csv", header=FALSE)
sumary(data)
# osztlyozs (nem szerinti)
#setwd("D:/Zoli2/Prog/R")
data = read.csv("crabdata.csv", header=FALSE)
summary(data)
# osztlyozs (nem szerinti)
#setwd("D:/Zoli2/Prog/R")
data = read.csv("crabdata.csv", header=FALSE)
summary(data)
library(nnet)
net = nnet(V7 ~ ., data=data, size=5)
result = predict(net, type="class")
t = table(data[["V7"]], result)
print(t)
cat("accuracy: ", 100 * sum(diag(t)) / sum(t), "%\n", sep="")
data <- read.csv("crabdata.csv", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V7 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 1:6], type="class")
t1 <- table(data[["V7"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 1:6], type="class")
t2 <- table(data[["V7"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("crabdata.csv", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V7 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 1:6], type="class")
t1 <- table(data[["V7"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 1:6], type="class")
t2 <- table(data[["V7"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("crabdata.csv", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V7 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 1:6], type="class")
t1 <- table(data[["V7"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 1:6], type="class")
t2 <- table(data[["V7"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("crabdata.csv", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V7 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 1:6], type="class")
t1 <- table(data[["V7"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 1:6], type="class")
t2 <- table(data[["V7"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("crabdata.csv", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V7 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 1:6], type="class")
t1 <- table(data[["V7"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 1:6], type="class")
t2 <- table(data[["V7"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("crabdata.csv", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V7 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 1:6], type="class")
t1 <- table(data[["V7"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 1:6], type="class")
t2 <- table(data[["V7"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("crabdata.csv", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V7 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 1:6], type="class")
t1 <- table(data[["V7"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 1:6], type="class")
t2 <- table(data[["V7"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("crabdata.csv", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V7 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 1:6], type="class")
t1 <- table(data[["V7"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 1:6], type="class")
t2 <- table(data[["V7"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("crabdata.csv", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V7 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 1:6], type="class")
t1 <- table(data[["V7"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 1:6], type="class")
t2 <- table(data[["V7"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("crabdata.csv", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V7 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 1:6], type="class")
t1 <- table(data[["V7"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 1:6], type="class")
t2 <- table(data[["V7"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("crabdata.csv", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V7 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 1:6], type="class")
t1 <- table(data[["V7"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 1:6], type="class")
t2 <- table(data[["V7"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("crabdata.csv", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V7 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 1:6], type="class")
t1 <- table(data[["V7"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 1:6], type="class")
t2 <- table(data[["V7"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("crabdata.csv", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V7 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 1:6], type="class")
t1 <- table(data[["V7"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 1:6], type="class")
t2 <- table(data[["V7"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("https://archive.ics.uci.edu/ml/machine-learning-databases/wine/wine.data", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V7 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 1:6], type="class")
t1 <- table(data[["V7"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 1:6], type="class")
t2 <- table(data[["V7"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("http://archive.ics.uci.edu/ml/machine-learning-databases/wine/wine.data", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V7 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 1:6], type="class")
t1 <- table(data[["V7"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 1:6], type="class")
t2 <- table(data[["V7"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("http://archive.ics.uci.edu/ml/machine-learning-databases/wine/wine.data", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V1 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 2:13], type="class")
t1 <- table(data[["V1"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 2:13], type="class")
t2 <- table(data[["V1"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("http://archive.ics.uci.edu/ml/machine-learning-databases/wine/wine.data", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V1 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 2:12], type="class")
t1 <- table(data[["V1"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 2:12], type="class")
t2 <- table(data[["V1"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("http://archive.ics.uci.edu/ml/machine-learning-databases/wine/wine.data", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V7 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 1:6], type="class")
t1 <- table(data[["V7"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 1:6], type="class")
t2 <- table(data[["V7"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("http://archive.ics.uci.edu/ml/machine-learning-databases/wine/wine.data", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V1 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 2:14], type="class")
t1 <- table(data[["V1"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 2:14], type="class")
t2 <- table(data[["V1"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("http://archive.ics.uci.edu/ml/machine-learning-databases/wine/wine.data", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V1 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 2:14], type="class")
t1 <- table(data[["V1"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 2:14], type="class")
t2 <- table(data[["V1"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("http://archive.ics.uci.edu/ml/machine-learning-databases/wine/wine.data", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V1 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 2:14], type="class")
t1 <- table(data[["V1"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 2:14], type="class")
t2 <- table(data[["V1"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
data <- read.csv("http://archive.ics.uci.edu/ml/machine-learning-databases/wine/wine.data", header=FALSE)
data <- transform(data, V1=factor(V1))
pairs(
data[2:6],
bg=c("red", "green", "blue")[data[[1]]],
pch=21,
cex=1.5
)
selected.rows <- sample(1:nrow(data), nrow(data) * 0.5)
library(nnet)
net <- nnet(
V1 ~ .,
data=data,
size=50,
maxit=300
)
t <- table(
y=predict(net, type="class"),
d=data[[1]]
)
print(t)
accuracy <- 100 * sum(diag(t)) / sum(t)
cat("accuracy: ", accuracy, "%\n", sep="")
f <- function(x) sin(pi * x) / (pi * x)
x <- runif(100, -5, 5)
d <- f(x)
d <- d + rnorm(length(d), sd=0.05)
plot(f, -5, 5)
points(x, d, col="blue")
library(nnet)
net <- nnet(x, d, linout=TRUE, size=10, maxit=200) # size:rtegek_szma
x2 <- seq(-5, 5, length=200)
y2 <-predict(net, matrix(x2))
plot(f, -5, 5)
title(main=expression(f(x) == frac(sin(pi * x), pi * x)))
points(x, d, col="blue", pch=1)
points(x2, y2, col="red", pch=4)
legend("topright",
c("Original function", "Training data", "Neural network"),
col=c("black", "blue", "red"),
lty=c("solid", "blank", "blank"),
pch=c(-1, 1,4)
)
data <- read.csv("http://archive.ics.uci.edu/ml/machine-learning-databases/wine/wine.data", header=FALSE)
summary(data)
library(nnet)
selected <- sample(1:nrow(data), 0.75 * nrow(data))
net <- nnet(V1 ~ ., data=data, size=5, subset=selected)
result1 <- predict(net, data[selected, 2:14], type="class")
t1 <- table(data[["V1"]][selected], result1)
print(t1)
cat("accuracy on training set: ", 100 * sum(diag(t1)) / sum(t1), "%\n", sep="")
result2 <- predict(net, data[-selected, 2:14], type="class")
t2 <- table(data[["V1"]][-selected], result2)
print(t2)
cat("accuracy on test set: ", 100 * sum(diag(t2)) / sum(t2), "%\n", sep="")
f <- function(x) sin(pi * x) / (pi * x)
x <- runif(100, -5, 5)
d <- f(x)
d <- d + rnorm(length(d), sd=0.05)
plot(f, -5, 5)
points(x, d, col="blue")
library(nnet)
net <- nnet(x, d, linout=TRUE, size=10, maxit=200) # size:rtegek_szma
x2 <- seq(-5, 5, length=200)
y2 <-predict(net, matrix(x2))
plot(f, -5, 5)
title(main=expression(f(x) == frac(sin(pi * x), pi * x)))
points(x, d, col="blue", pch=1)
points(x2, y2, col="red", pch=4)
legend("topright",
c("Original function", "Training data", "Neural network"),
col=c("black", "blue", "red"),
lty=c("solid", "blank", "blank"),
pch=c(-1, 1,4)
)
q()
