dwdm-prac1-full

#loading data set
data = iris
#details of data set
summary(data)
#display top 5 rows
head(data)

#writing a function to normalise a column
normalise = function(x){
  return ((x - min(x))/(max(x)-min(x)))
}

#yanking columns from the data set and normalizing them
data$Sepal.Length = normalise(data$Sepal.Length)
data$Sepal.Width = normalise(data$Sepal.Width)
data$Petal.Length = normalise(data$Petal.Length)
data$Petal.Width = normalise(data$Petal.Width)

#randomizing data to make a data set for training
ind = sample(1:nrow(data), size = 0.9*nrow(data), replace = FALSE )
training_data = data[ind,]

#creating testing data (500 IQ move)
test_data = data[-ind,]

#creating label for later verification (5th col of table)
test_data_label = test_data[,5]

#removing 5th column from test_data
test_data = test_data[-5]

#creating training data label
training_data_label = training_data[,5]
training_data = training_data[,-5]

#load package
library(class)
library(caret)
library(ggplot2)

#executing(implementing?) KNN algorithm
model = knn(training_data, test_data, training_data_label, k=11)

#comparing the model and the test data labels
model
test_data_label

#evaluating the performance of the model
confusionMatrix(model, test_data_label)

#there is another

#resetting training data
training_data = data[ind,]

#another model
model = train(Species ~., data = training_data, method = 'knn')

#another prediction
prediction = predict(model, test_data)

#evaluating the performance of the new model
confusionMatrix(prediction, test_data_label)