Applied Supervised Learning with R

Applied Supervised Learning with R

By : Karthik Ramasubramanian, Jojo Moolayil

Buy this Book

Applied Supervised Learning with R

By: Karthik Ramasubramanian, Jojo Moolayil

Buy this Book

Overview of this book

R provides excellent visualization features that are essential for exploring data before using it in automated learning. Applied Supervised Learning with R helps you cover the complete process of employing R to develop applications using supervised machine learning algorithms for your business needs. The book starts by helping you develop your analytical thinking to create a problem statement using business inputs and domain research. You will then learn different evaluation metrics that compare various algorithms, and later progress to using these metrics to select the best algorithm for your problem. After finalizing the algorithm you want to use, you will study the hyperparameter optimization technique to fine-tune your set of optimal parameters. The book demonstrates how you can add different regularization terms to avoid overfitting your model. By the end of this book, you will have gained the advanced skills you need for modeling a supervised machine learning algorithm that precisely fulfills your business needs.

Applied Supervised Learning with R

Preface

Free Chapter

R for Advanced Analytics

Introduction

Working with Real-World Datasets

Reading Data from Various Data Formats

Write R Markdown Files for Code Reproducibility

Data Structures in R

DataFrame

Data Processing and Transformation

The Apply Family of Functions

Boxplot

Summary

Exploratory Analysis of Data

Introduction

Defining the Problem Statement

Understanding the Science Behind EDA

Exploratory Data Analysis

Univariate Analysis

Exploring Categorical Features

Bivariate Analysis

Studying the Relationship between Two Numeric Variables

Studying the Relationship between a Categorical and a Numeric Variable

Studying the Relationship Between Two Categorical Variables

Multivariate Analysis

Validating Insights Using Statistical Tests

Categorical Dependent and Numeric/Continuous Independent Variables

Categorical Dependent and Categorical Independent Variables

Summary

Introduction to Supervised Learning

Introduction

Summary of the Beijing PM2.5 Dataset

Regression and Classification Problems

Machine Learning Workflow

Regression

Exploratory Data Analysis (EDA)

Classification

Evaluation Metrics

Summary

Regression

Introduction

Linear Regression

Model Diagnostics

Residual versus Fitted Plot

Normal Q-Q Plot

Scale-Location Plot

Residual versus Leverage

Improving the Model

Quantile Regression

Polynomial Regression

Ridge Regression

LASSO Regression

Elastic Net Regression

Poisson Regression

Cox Proportional-Hazards Regression Model

NCCTG Lung Cancer Data

Summary

Classification

Introduction

Getting Started with the Use Case

Classification Techniques for Supervised Learning

Logistic Regression

How Does Logistic Regression Work?

Evaluating Classification Models

What Metric Should You Choose?

Evaluating Logistic Regression

Decision Trees

XGBoost

Deep Neural Networks

Choosing the Right Model for Your Use Case

Summary

Feature Selection and Dimensionality Reduction

Highly Correlated Variables

Feature Reduction

Variable Clustering

Linear Discriminant Analysis for Feature Reduction

Summary

Model Improvements

Introduction

Bias-Variance Trade-off

Underfitting and Overfitting

Defining a Sample Use Case

Cross-Validation

Holdout Approach/Validation

K-Fold Cross-Validation

Hold-One-Out Validation

Hyperparameter Optimization

Grid Search Optimization

Random Search Optimization

Bayesian Optimization

Summary

Model Deployment

Introduction

What is an API?

Introduction to plumber

A Brief History of the Pre-Docker Era

Docker

Amazon Web Services

Introducing AWS SageMaker

What is Amazon Lambda?

What is Amazon API Gateway?

Building Serverless ML Applications

Deleting All Cloud Resources to Stop Billing

Summary

Capstone Project - Based on Research Papers

Introduction

Exploring Research Work

The mlr Package

Problem Design from the Research Paper

Features in Scene Dataset

Implementing Multilabel Classifier Using the mlr and OpenML Packages

Constructing a Learner

Predictions

Summary

Appendix

Chapter 1: R for Advanced Analytics

Chapter 2: Exploratory Analysis of Data

Chapter 3: Introduction to Supervised Learning

Chapter 4: Regression

Chapter 5: Classification

Chapter 6: Feature Selection and Dimensionality Reduction

Chapter 7: Model Improvements

Chapter 8: Model Deployment

Chapter 9: Capstone Project - Based on Research Papers

Customer Reviews

5 star

4 star

3 star

2 star

1 star

Chapter 4: Regression

Activity 7: Printing Various Attributes Using Model Object Without Using the summary Function

First, print the coefficient values using the following command. Make sure the output is like the output of the summary function using the coefficients option. The coefficients are the fitted values from the model that uses the OLS algorithm:
```
multiple_PM25_linear_model$coefficients
```
The output is as follows:
```
(Intercept)        DEWP        TEMP         Iws 
161.1512066   4.3841960  -5.1335111  -0.2743375
```
Find the residual value (difference) of the predicted and actual values of PM2.5, which should be as small as possible. Residual reflects how far the fitted values using the coefficients are from the actual value.
```
multiple_PM25_linear_model$residuals
```
The output is as follows:
```
25            26            27            28 
  17.95294914   32.81291348   21.38677872   26.34105878 
           29            30            31            32 
```

Next, find the fitted values that should be closer to the actual PM2.5 for the best model. Using the coefficients, we can compute the fitted values:

multiple_PM25_linear_model$fitted.values

The output is as follows:

25         26         27         28         29 
111.047051 115.187087 137.613221 154.658941 154.414781 
        30         31         32         33         34

Find the R-Squared values. They should look the same as the one you obtained in the output of the summary function next to the text Multiple R-squared. R-Square helps in evaluating the model performance. If the value is closer to 1, the better the model is:
```
summary(multiple_PM25_linear_model)$r.squared
```
The output is as follows:
```
[1] 0.2159579
```
Find the F-Statistic values. Make sure the output should look same as the one you obtained in the output of the summary function next to the text F-Statistics. This will tell you if your model fits better than just using the mean of the target variable. In many practical applications, F-Statistic is used along with p-values:
```
summary(multiple_PM25_linear_model)$fstatistic
```
The output is as follows:
```
    value     numdf     dendf 
 3833.506     3.000 41753.000 
```
Finally, find the coefficient p-values and make sure the values should look the same as the one you obtained in the output of the summary function under Coefficients for each variable. It will be present under the column titled Pr(>|t|):. If the value is less than 0.05, the variable is statistically significant in predicting the target variable:
```
summary(multiple_PM25_linear_model)$coefficients[,4]
```
The output is as follows:
```
  (Intercept)          DEWP          TEMP           Iws 
 0.000000e+00  0.000000e+00  0.000000e+00 4.279601e-224
```

The attributes of a model are equally essential to understand, especially in linear regression than to obtain the prediction. They help in interpreting the model well and connect the problem to its real use case.

Applied Supervised Learning with R

By : Karthik Ramasubramanian, Jojo Moolayil

Applied Supervised Learning with R

By: Karthik Ramasubramanian, Jojo Moolayil

Overview of this book

Related Content you might be interested in

Current Title:

Applied Supervised Learning with R

Practical Machine Learning with R

Mastering Machine Learning with R

Regression Analysis with R

Chapter 4: Regression

Activity 7: Printing Various Attributes Using Model Object Without Using the summary Function