Difference between revisions of "Machine-Learning-using-R/C2/Introduction-to-Machine-Learning-in-R/English"

Opening Slide

Learning Objectives

• Machine Learning
• Supervised and Unsupervised Learning
• Workflow of ML CLassifier Algorithm
• Visualizing Feature Space
• Constructing a dummy classifier
• Evaluation of the chosen dummy classifier

System Specifications

• Windows 11
• R version 4.3.0
• RStudio version 2023.06.1

It is recommended to install R version 4.2.0 or higher.

Prerequisites

https://spoken-tutorial.org

• Basic programming in R.
• To use GGPlot2 and dplyr package.

If not, please access the relevant tutorials on this website.

Machine Learning

• ML enables computers to learn from data.
• ML algorithms automate the learning process from data through patterns.
• Their primary role is prediction, classification or clustering of data.
• ML algorithms are applied in several applications.
• For example Natural Language Processing, Image and speech recognition, etc.

Types of Machine Learning

• Supervised learning: Prediction and Classification,
• Unsupervised learning: Clustering,
• Semi-supervised learning
• Reinforcement learning.

In this series, we will focus on Supervised and Unsupervised learning algorithms.

Supervised and Unsupervised Learning

• ML algorithms predict labels for unseen features
• They predict based on given features and labels of data.

Unsupervised learning: Unlabeled data

• ML algorithms develop a mechanism to group similar features into clusters.
• And label them for future analysis.

Classification and Regression

• Supervised learning consists of Regression and Classification.
• Regression is applied to predict and learn continuous-valued responses from features.
• Regression techniques include Linear, Spline, Ridge, Lasso, and others.
• Classification is applied to predict the class of a discrete (labeled) response from features.
• Classification techniques include Logistic Regression, Decision Tree, SVM, and others.

Workflow of an ML Classifier algorithm

• Feature Space: Collection of all possible values of the features.
• A classification algorithm partitions the feature space into a number of classes.
• Data is split into training and testing sets to learn and evaluate the algorithm.
• The model learns from the training data to create partitions of feature space.
• The model is evaluated on the test dataset through performance metrics.

Dataset

For more information on Raisin data please refer to Additional Reading Material on this tutorial page.

Download Files

Please download these files from the Code files link of this tutorial.

Make a copy and then use them while practicing.

point to Intro.R and the folder Introduction.

Point to the MLProject folder on the Desktop.

This folder is located in the MLProject folder on my Desktop.

I have also set the Introduction folder as my working Directory.

In this tutorial, we will introduce classification on the raisin dataset.

Point to Intro.R in RStudio.

Script Intro.R opens in RStudio.

Highlight the command library(readxl)

Highlight the command library(caret)

Highlight the command library(ggplot2)

#install.packages(“package_name”)

Point to the command.

Select and run these commands to import the packages.

We will use the readxl package to load the excel file of our Raisin Dataset.

We will use the caret package to create the confusion matrix.

The ggplot2 package will be used to create the decision boundary plot.

Please ensure that all the packages are installed correctly.

As I have already installed the packages, I have imported them directly.

Highlight the command

data<- read_xlsx("Raisin.xlsx")

Drag boundary to see the Environment tab clearly.

In the Environment tab below Data, you will see the data variable.

Click on data to load the dataset in the Source window.

Click on Intro.R in the Source window and close the tab.

data<-data[c("minorAL",ecc,"class")]

data$class <- factor(data$class)

Select the commands and click the Run button

We now select three columns from data.

2 columns ("minorAL", "ecc") are chosen as features.

The class column is chosen as a target variable.

We convert the target variable data$class to a factor.

Select and run the commands.

Click on data.

range_ecc <- range(data$ecc)

range_minor_al <- range(data$minorAL)

Highlight the command

range_ecc <- range(data$ecc)

Select and run the commands.

Drag boundary to see the environment tab clearly.

The minimum and maximum value of the minor_al and ecc are shown in their range variables

Y <- seq(min(data$ecc), max(data$ecc), length.out = 100)

feature <- expand.grid(minorAL = X, ecc = Y)

In the Source window type these commands

X <- seq(min(data$minorAL), max(data$minorAL), length.out = 100)

Y <- seq(min(data$ecc), max(data$ecc), length.out = 100)

HIghlight

feature <- expand.grid(minorAL = X, ecc = Y)

This command creates a cartesian product of the two features to create a feature space.

Select and run the commands.

geom_point(aes(color = class), size = 2) +

scale_fill_manual(values = c("#ffff46", "#FF46e9")) +

scale_color_manual(values = c("red", "blue")) +

labs(title = "Feature Space") +

theme_minimal()

In the Source window type these commands

geom_point(aes(color = class), size = 2) +

scale_fill_manual(values = c("#ffff46", "#FF46e9")) +

scale_color_manual(values = c("red", "blue")) +

labs(title = "Feature Space") +

theme_minimal()

Select and run the commands.

set.seed(1)

index_split<- sample(1:nrow(data),size=0.7*nrow(data),replace=FALSE)

Click on Intro.R in the Source window, and type these commands.

set.seed(1)

Highlight the command

index_split<- sample(1:nrow(data),size=0.7*nrow(data),replace=FALSE)

Select the commands and run them.

train_data <- data[index_split, ]

test_data <- data[-c(index_split), ]

train_data <- data[index_split, ]

Highlight the command

test_data <- data[-c(index_split), ]

This creates testing data, consisting of 270 unique rows.

Point to the sets in the Environment Tab

Click the test_data and train_data

Select the commands and run them.

The data sets are shown in the Environment tab.

Drag boundary to see the Environment tab clearly

Click on test_data and train_data to load them in the Source window.

To begin with, one might construct a heuristic line to build the classifier.

fit = function(x)((x * (-0.0021)) + 1.445)

model_predict <- function(x){

factor(ifelse(x$ecc < fit(x$minorAL), "Kecimen", "Besni"))

}

fit = function(x)((x * (-0.0021)) + 1.445)

Highlight the command

model_predict <- function(x){

factor(ifelse(x$ecc < fit(x$minorAL), "Kecimen", "Besni"))

}

Click Save and Click Run buttons.

For that we will define a line to partition the data as a dummy classifier.

It doesn’t involve training data so performance may be poor.

We define a function that separates data points belonging to either side of the line.

Click Save.

Select and run the commands.

feature$classnum <- as.numeric(feature$class)

In the Source window type these commands

feature$class <- model_predict(feature)

Highlight

feature$classnum <- as.numeric(feature$class)

This command will use the line created to predict the class of every point in the grid of feature space.

This command encodes the class string labels into numbers suitable for plotting

Select and run the commands.

Point to the data in the Source window.

Click on feature in the Environment tab.

The feature set with the predicted classes loads in the source window.

geom_raster(data= feature, aes(x=minorAL, y=ecc, fill = class),alpha=0.3) +

geom_point(data = data, aes(x = minorAL, y = ecc, color = class), size = 2) +

geom_abline(slope = -0.0021, intercept = 1.445, size = 1.2)+

scale_fill_manual(values = c("#ffff46", "#FF46e9")) +

scale_color_manual(values = c("red", "blue")) +

labs(title = "Data Boundary") +

theme_minimal()

ggplot() +

geom_raster(data= feature, aes(x=minorAL, y=ecc, fill = class),alpha=0.3) +

geom_point(data = data, aes(x = minorAL, y = ecc, color = class), size = 2) +

geom_abline(slope = -0.0021, intercept = 1.445, size = 1.2)+

scale_fill_manual(values = c("#ffff46", "#FF46e9")) +

scale_color_manual(values = c("red", "blue")) +

labs(title = "Data Boundary") +

theme_minimal()

We are visualising the feature space and the partition line using GGPlot2.

Select and run the commands.

Overall plot shows that the chosen line approximately separates the training data classes.

prediction_test = model_predict(test_data)

In the Source window type this command

prediction_test = model_predict(test_data)

We predict the classes from testing data and store it in the prediction_test variable.

Select and run the command.

test_confusion_matrix <- confusionMatrix(test_data$class,prediction_test)

In the Source window, type the command

test_confusion_matrix <- confusionMatrix(test_data$class,prediction_test)

Click on Save and Run buttons.

Select and run the command.

test_confusion_matrix$overall["Accuracy"]

Select and run the command

Accuray

0.6962963

Let us now view the confusion matrix of the testing dataset

test_confusion_matrix$table

test_confusion_matrix$table

Click on Save and Run buttons.

The output is seen in the console window

Reference

Prediction Besni Kecimen

Besni 50 82

Kecimen 0 138

Observe that:

0 samples of class Besni have been incorrectly classified.

82 samples of class Kecimen have been incorrectly classified.

We can see that our partition line is skewed.

We can choose a complicated partition to reduce train misclassification error.

But there will be no control on test data.

We can aim to choose a classifier which is simple with a smaller test misclassification error.

Let us summarize.

Summary

• Machine Learning
• Classification and Regression Problems
• Workflow of an ML Classifier Algorithm
• Visualizing Feature Space
• Constructing a dummy classifier
• Evaluation of an ML algorithm

Assignment

• Use a vertical line as a classifier to partition the feature space.
• Plot the decision boundary for the same.
• Evaluate the classifier on the test dataset

About the Spoken Tutorial Project

Please download and watch it.

Spoken Tutorial Workshops

Please contact us.

Spoken Tutorial Forum to answer questions

Do you have questions in THIS Spoken Tutorial?

Choose the minute and second where you have the question.

Explain your question briefly.

Someone from our team will answer them.

Please visit this site.

Forum to answer questions

Please visit the forum given in the link.

R Activities

We give certificates to those who do this.

For more details, please visit the website.

Acknowledgment

Thank You

Thank you for joining.