Difference between revisions of "Machine-Learning-using-R/C2/Logistic-Regression-in-R/English"

Opening Slide

Learning Objectives

• Logistic Regression
• Assumptions of Logistic Regression
• Advantages of Logistic Regression
• Implementation of Logistic Regression in R using Raisin dataset.
• Model Evaluation.
• Visualization of the model Decision Boundary
• Limitations of Logistic Regression

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

• Basic programming in R.
• Basics of Machine Learning.

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

Logistic Regression

It models the probability of success for the explanatory variable.

• It predicts the probability, unlike the response in linear regression.
• The predicted probability is used as a classifier.
• The probability of success is modeled using the logit or (log odds) function.
• It is a linear classifier, as the logistic regression model has a linear logit.
• It is often used when the response variable is categorical.

Assumptions of Logistic Regression

• The distribution of the dependent variable is Bernoulli.
• The data records are independent.

Advantages of Logistic Regression

• It provides estimates of regression coefficients along with their standard errors.
• It also provides the predicted probability which in turn is used as a classifier.
• It doesn’t need explanatory variables to be necessarily continuous.
• In this sense, it is a more general classifier than LDA and QDA.

Implementation Of Logistic Regression

The additional reading material has more details on the Raisin dataset.

Please refer to it.

Download Files

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

Make a copy and then use them while practicing.

Highlight LogisticRegression.R

Logistic Regression folder.

This folder is located in the MLProject folder.

I have also set the Logistic Regression folder as my Working Directory.

Let’s create a Logistic Regression classifier model on the raisin dataset.

Point to LogisticRegression.R in RStudio.

For this, click on the script LogisticRegression.R.

Script LogisticRegression.R opens in RStudio.

Highlight the commands

library(readxl)

library(caret)

library(VGAM)

library(ggplot2)

library(dplyr)

#install.packages(“package_name”)

Point to the command.

The VGAM package contains the glm() function required to create our classifier.

As I have already installed the packages.

I have directly imported them.

Highlight

data <- read_xlsx("Raisin_Dataset.xlsx")

data[c("minorAL",”ecc”,"class")]

data$class <- factor(data$class)

Highlight the commands.

They will also prepare the dataset for model building.

Select and run the commands.

Click on data on the Environment tab.

Click on data in the Environment tab.

It loads the modified dataset in the Source window.

set.seed(1)

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

train <- data[trainIndex, ]

test <- data[-trainIndex, ]

In the Source window type these commands.

set.seed(1)

Highlight

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

Highlight

train <- data[trainIndex, ]

Highlight

test <- data[-trainIndex, ]

Click on Save and Run buttons.

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

Select the commands and run them.

Logistic_model <- glm(class ~ ., data = train, family = "binomial")

summary(Logistic_model)$coef

In the Source window type these commands

Highlight class ~ .

Highlight family = binomial

Highlight train

Logistic regression is among the class of models that it fits.

This is the formula for our model.

We try to predict target variable class based on minorAL and ecc features.

This ensures that our model predicts the probability for 2 classes.

It ensures that, out of all the models in glm, the logistic regression model is fit.

This is the data used to train our model.

Select the commands and run them.

The output is shown in the console window.

Highlight Coefficients

Highlight Pr(>|z|)

Coefficients denote the coefficients of the logit function.

That means the log-odds of class change by -0.04 for every unit change in minorAL.

The lower p-values suggest that the effects are statistically significant.

Predicted.prob <- predict(Logistic_model, test, type="response")

View(Predicted.prob)

In the Source window type these commands

Predicted.prob <- predict(Logistic_model, test, type="response")

Highlight

Type = “response”

This command ensures the outcome is a probability.

Select and run the commands

Value

Predicted.prob stores the predicted probability of each observation belonging to a certain class.

predicted.classes <- factor(ifelse(predicted.prob > 0.5, "Kecimen", "Besni"))

If the probability is greater than 0.5 then Kecimen class otherwise Besni Class is chosen

We also convert the output to a factor datatype to fit in the Confusion matrix function.

Select and run the commands

Point to the confusion in the Environment Tab

Highlight the attribute

table

List is created from the actual and predicted class labels.

And it is stored in the confusion_matrix variable.

It helps to assess the classification model's performance and accuracy.

Select and run these commands

tab <- confusion_matrix$table

tab = as.data.frame(tab)

tab$Prediction <- factor(tab$Prediction, levels = rev(levels(tab$Prediction)))

tab <- tab %>%

rename(Actual = Reference) %>%

mutate(cor = if_else(Actual == Prediction, 1,0))

tab$cor <- as.factor(tab$cor)

ggplot(tab, aes(Actual,Prediction)) +

geom_tile(aes(fill= cor),alpha = 0.4) + geom_text(aes(label=Freq)) +

scale_fill_manual(values = c("red","green")) +

theme_light() +

theme(legend.position = "None",

line = element_blank()) +

scale_x_discrete(position = "top")

}

Highlight the command

tab <- confusion_matrix$table

Highlight the command

tab <- confusion_matrix$table

tab = as.data.frame(tab)

tab$Prediction <- factor(tab$Prediction, levels = rev(levels(tab$Prediction)))

tab <- tab %>%

rename(Actual = Reference) %>%

mutate(cor = if_else(Actual == Prediction, 1,0))

tab$cor <- as.factor(tab$cor)

Highlight the command

ggplot(tab, aes(Actual,Prediction)) +

geom_tile(aes(fill= cor),alpha = 0.4) + geom_text(aes(label=Freq)) +

scale_fill_manual(values = c("red","green")) +

theme_light() +

theme(legend.position = "None",

line = element_blank()) +

scale_x_discrete(position = "top")

}

It fetches the confusion matrix table from the list.

It creates a data frame from the table which is suitable for plotting using GGPlot2.

It plots the confusion matrix using the data frame created.

It represents correct and incorrect predictions using different colors.

Select and run the commands

plot_confusion_matrix(confusion)

In the Source window type this command

Highlight the command

plot_confusion_matrix(confusion)

Click on Save and Run buttons.

We use the plot_confusion_matrix() function to generate a visual plot of the confusion matrix list created.

Select and run the command

The output is seen in the plot window

We observe that:

21 misclassifications of Besni Class.

13 misclassifications of Kecimen class.

grid <- expand.grid(minorAL = seq(min(data$minorAL), max(data$minorAL), length = 500),

ecc = seq(min(data$ecc), max(data$ecc), length = 500))

grid$prob <- predict(model, newdata = grid, type = "response")

grid$class <- ifelse(grid$prob > 0.5, 'Kecimen', 'Besni')

grid$classnum <- as.numeric(as.factor(grid$class))

In the Source window type these commands

grid <- expand.grid(minorAL = seq(min(data$minorAL), max(data$minorAL), length = 500),

ecc = seq(min(data$ecc), max(data$ecc), length = 500))

grid$prob <- predict(model, newdata = grid, type = "response")

grid$class <- ifelse(grid$prob > 0.5, 'Kecimen', 'Besni')

grid$classnum <- as.numeric(as.factor(grid$class))

Then, it uses the Logistics Regression model to predict the probability of each point in this grid, storing these predictions as a new column 'prob' in the grid dataframe.

It converts the predicted probabilities of the points into classes.

If the probability exceeds 0.5 then Kecimen class otherwise Besni Class is chosen.

The prediced classes are stored in ‘class’ column of grid data frame.

The as.numeric function encodes the predicted classes string labels into numeric values.

Select and run the commands

Click on grid in the Environment tab to load the generated data in the Source window.

ggplot() +

geom_raster(data = grid, aes(x = minorAL, y = ecc, fill = class), alpha = 0.4) +

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

geom_contour(data = grid, aes(x = minorAL, y = ecc, z = classnum),

colour = "black", linewidth = 0.7) +

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

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

labs(x = "MinorAL", y = "ecc", title = "Logistic Regression Decision Boundary") +

theme_minimal()

ggplot() +

geom_raster(data = grid, aes(x = minorAL, y = ecc, fill = class), alpha = 0.4) +

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

geom_contour(data = grid, aes(x = minorAL, y = ecc, z = classnum),

colour = "black", linewidth = 0.7) +

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

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

labs(x = "MinorAL", y = "ecc", title = "Logistic Regression Decision Boundary") +

theme_minimal()

It plots the grid points with colors indicating the predicted classes.

The overall plot provides a visual representation of the decision boundary and the distribution of training data points of the model.

Select and run these commands.

Drag boundaries to see the plot window clearly.

The line separates the data points clearly.

Limitations of Logistic Regression

• It’s sensitive to outliers which can affect the accuracy of the classifier.
• It can perform poorly in the presence of multicollinearity among explanatory variables.

Summary

• Logistic Regression
• Assumptions of Logistic Regression
• Advantages of Logistic Regression
• Implementation of Logistic Regression in R using Raisin dataset.
• Model Evaluation.
• Visualization of the model Decision Boundary
• Limitations of Logistic Regression

Assignment

• Apply logistic regression on the Wine dataset.
• This dataset can be found in the HDclassif package.
• Install the package and import the dataset using the data() command.
• Measure the accuracy of the model

About the Spoken Tutorial Project

Spoken Tutorial Workshops

Please contact us.

Spoken Tutorial Forum to answer questions

Forum to answer questions

Please visit the forum given in the link.

Textbook Companion

We give certificates to those who do this.

For more details, please visit these sites.

Acknowledgment

Thank You

Thank you for joining.