Introduction to Machine Learning in the Cloud with Python: Concepts and Practices

Foreword
Preface
About the Book
Contents
Acronyms
Part I: Concepts
	Chapter 1: Machine Learning Concepts
		1.1 Terminology
		1.2 What Is Machine Learning?
			1.2.1 Mitchell´s Notion of Machine Learning
		1.3 What Does Learning Mean for a Computer?
		1.4 Difference Between ML and Traditional Programming
		1.5 How Do Machines Learn?
		1.6 Steps to Apply ML
		1.7 Paradigms of Learning
			1.7.1 Supervised Machine Learning
			1.7.2 Unsupervised Machine Learning
			1.7.3 Reinforcement Machine Learning
				1.7.3.1 Types of Problems in Machine Learning
		1.8 Machine Learning in Practice
		1.9 Why Use Machine Learning?
		1.10 Why Machine Learning Now?
		1.11 Classical Tasks for Machine Learning
		1.12 Applications of Machine Learning
			1.12.1 Applications in Our Daily Life
		1.13 ML Computing Needs
		1.14 Machine Learning in the Cloud
		1.15 Tools Used in Machine Learning
		1.16 Points to Ponder
		References
	Chapter 2: Machine Learning Algorithms
		2.1 Why Choose Machine Learning?
		2.2 Supervised Machine Learning Algorithms
			2.2.1 Regression
			2.2.2 Classification
			2.2.3 Machine Learning Algorithms: Supervised Learning
			2.2.4 Machine Learning Algorithms: Unsupervised Learning
				2.2.4.1 Clustering
				2.2.4.2 Dimension Reduction
				2.2.4.3 Anomaly Detection
			2.2.5 Machine Learning Algorithms That Use Unsupervised Learning
		2.3 Considerations in Choosing an Algorithm
		2.4 What Are the Most Common and Popular Machine Learning Algorithms?
			2.4.1 Linear Regression
			2.4.2 Two Types of Linear Regression
				2.4.2.1 Simple Linear Regression
				2.4.2.2 Multiple Linear Regression
				2.4.2.3 Assumptions of Linear Regression
				2.4.2.4 Advantages
				2.4.2.5 Disadvantages
				2.4.2.6 Sample Python Code for Linear Regression
			2.4.3 K-Nearest Neighbors (KNN)
				2.4.3.1 Assumptions
				2.4.3.2 How Does KNN Algorithm Works?
				2.4.3.3 Advantages
					Sample Python Code for KNN Algorithm
			2.4.4 Logistic Regression
				2.4.4.1 Types of Logistic Regression
				2.4.4.2 Assumptions
				2.4.4.3 Advantages
				2.4.4.4 Disadvantages
					Sample Python Code for Implementing Logistic Regression
			2.4.5 Naïve Bayes Classifier Algorithm
				2.4.5.1 Additive Smoothing
				2.4.5.2 Types of Naïve Bayes Model
				2.4.5.3 Assumptions
				2.4.5.4 How Naïve Bayes Algorithm Works?
				2.4.5.5 Advantages
				2.4.5.6 Disadvantages
					Sample Python Code for Naïve Bayes Model
			2.4.6 Support Vector Machine Algorithm
				2.4.6.1 Assumptions
				2.4.6.2 Types of SVM
				2.4.6.3 Advantages
				2.4.6.4 Disadvantages
					Sample Python Code for SVM
			2.4.7 Decision Trees
				2.4.7.1 Information Gain
				2.4.7.2 Gini Index
				2.4.7.3 Decision Tree Terminology
				2.4.7.4 Assumptions
				2.4.7.5 How Does the Decision Tree Classifier Work?
				2.4.7.6 Advantages
				2.4.7.7 Disadvantages
					Sample Python Code for Decision Tree
			2.4.8 Ensemble Learning
				2.4.8.1 Types of Ensemble Learning
					Boosting
					Bootstrap Aggregation (Bagging)
			2.4.9 Random Forests
				2.4.9.1 Assumptions
				2.4.9.2 How Does Random Forest Algorithm Work?
				2.4.9.3 Advantages
				2.4.9.4 Disadvantages
					Sample Python Code for Random Forest
			2.4.10 K-Means Clustering Algorithm
				2.4.10.1 Assumptions
				2.4.10.2 How Does K-Means Algorithm Work?
				2.4.10.3 Convergence Criterion
				2.4.10.4 Advantages
				2.4.10.5 Disadvantages
					Sample Python Code for K-Means
			2.4.11 Artificial Neural Networks
				2.4.11.1 Advantages
				2.4.11.2 Disadvantages
		2.5 Usage of ML Algorithms
		2.6 Performance Metrics of ML Algorithms
			2.6.1 Testing Data
			2.6.2 Performance Metrics for Classification Models
				2.6.2.1 Confusion Matrix
				2.6.2.2 Regression Metrics
		2.7 Most Popular Machine Learning Software Tools
		2.8 Machine Learning Platforms
			2.8.1 Alteryx Analytics
			2.8.2 H2O.ai
			2.8.3 KNIME Analytics Platform
			2.8.4 RapidMiner
			2.8.5 Databricks Unified Analytics Platform
			2.8.6 Microsoft´s Azure Machine Learning Studio
			2.8.7 Google´s Analytics Platform
			2.8.8 IBM Watson
			2.8.9 Amazon Web Services (AWS)
		2.9 Points to Ponder
		References
	Chapter 3: Deep Learning and Cloud Computing
		3.1 Deep Learning (DL)
		3.2 Historical Trends
		3.3 How Do Deep Learning Algorithm Learn?
			3.3.1 Activation Functions
		3.4 Architectures
			3.4.1 Deep Neural Network (DNN)
			3.4.2 Recurrent Neural Network (RNN)
			3.4.3 Convolutional Neural Networks (CNN)
		3.5 Choosing a Network
		3.6 Deep Learning Development Flow
		3.7 What Is Deep About Deep Learning?
		3.8 Data Used for Deep Learning
		3.9 Difference Between Machine Learning and Deep Learning
		3.10 Why Deep Learning Became Popular Now?
		3.11 Should You Always Use Deep Learning Instead of Machine Learning?
		3.12 Why Is Deep Learning Important?
		3.13 What Are the Drawbacks of Deep Learning?
		3.14 Which Deep Learning Software Frameworks Are Available?
		3.15 Classical Problems of Deep Learning Solves
			3.15.1 Image Classification
			3.15.2 Natural Language Processing
		3.16 The Future of Deep Learning
		3.17 Points to Ponder
		References
	Chapter 4: Cloud Computing Concepts
		4.1 Roots of Cloud Computing
		4.2 Key Characteristics of Cloud Computing
		4.3 Various Cloud Stakeholders
		4.4 Pain Points in Cloud Computing
		4.5 AI and ML in Cloud
		4.6 Expanding Cloud Reach
		4.7 Future Trends
		4.8 Summary
		4.9 Points to Ponder
		References
Part II: Practices
	Chapter 5: Practical Aspects in Machine Learning
		5.1 Preprocessing Data
		5.2 Challenges in Data Preparation
		5.3 When to Use Data Preprocessing?
		5.4 Framework for Data Preparation Techniques
			5.4.1 Data Preparation
			5.4.2 Data Selection (Feature Selection)
			5.4.3 Data Preprocessing
			5.4.4 Data Cleaning
			5.4.5 Insufficient Data
			5.4.6 Non-representative Data
			5.4.7 Substandard Data
			5.4.8 Data Transformation
			5.4.9 Handling Missing Values
		5.5 Modification of Categorical or Text Values to Numerical Values
		5.6 Feature Scaling
			5.6.1 Techniques of Feature Scaling
				5.6.1.1 Feature Scaling: Standardization
				5.6.1.2 Feature Scaling: Normalization (Min-Max Normalization)
		5.7 Inconsistent Values
		5.8 Duplicated Values
		5.9 Feature Aggregation
		5.10 Feature Sampling
			5.10.1 Sampling Without Replacement
			5.10.2 Sampling with Replacement
		5.11 Multicollinearity and Its Impact
		5.12 Feature Selection
			5.12.1 Importance of Feature Selection
			5.12.2 How Many Features to Have in the Model?
			5.12.3 Types of Feature Selection
				5.12.3.1 Filter Method
				5.12.3.2 Wrapper Methods
				5.12.3.3 Embedded Methods (Fig. 5.7)
					LASSO Regression
					Ridge Regression
		5.13 Dimensionality Reduction
			5.13.1 Principal Component Analysis (PCA)
			5.13.2 Linear Discriminant Analysis
			5.13.3 t-Distributed Stochastic Neighbor Embedding (t-SNE)
		5.14 Dealing with Imbalanced Data
			5.14.1 Use the Right Evaluation Metrics
			5.14.2 Sampling-Based Approaches
			5.14.3 Algorithm Based Approach
		5.15 Points to Ponder
		References
	Chapter 6: Information Security and Cloud Computing
		6.1 Information Security Background and Context
		6.2 Privacy Issues
		6.3 Security Concerns of Cloud Operating Models
		6.4 Secure Transmissions, Storage, and Computation
		6.5 A Few Key Challenges Related to Cloud Computing and Virtualization
		6.6 Security Practices for Cloud Computing
		6.7 Role of ML for Cybersecurity
		6.8 Summary
		6.9 Points to Ponder
		References
	Chapter 7: Examples of Analytics in the Cloud
		7.1 Background
		7.2 Analytics Services in the Cloud
		7.3 Introduction to MapReduce
		7.4 Introduction to Hadoop
		7.5 Examples of Cloud-Based ML
			7.5.1 Cloud Security Monitoring Using AWS
			7.5.2 Greener Energy Future with ML in GCP
			7.5.3 Monorail Monitoring in Azure
			7.5.4 Detecting Online Hate Speech Using NLP
		7.6 Future Possibilities
		7.7 Summary
		7.8 Points to Ponder
		References
	Chapter 8: Health Care in the Cloud: A Few Case Studies
		8.1 Introduction
		8.2 Existing TCD Solution
		8.3 Trail of Bubbles
		8.4 Moving Data to the Cloud
		8.5 A Reader in the Cloud
		8.6 Cloud-Based Collaborative Tools
		8.7 Multi-Cloud Solutions
		8.8 UCSD Antibiogram: Using Unclassifiable Data
		8.9 Next Steps
		8.10 Summary
		8.11 Points to Ponder
		References
	Chapter 9: Trends in Hardware-Based AL and ML
		9.1 Revisiting the History of AI
		9.2 Current Limitations of AI and ML
		9.3 Emergence of AI Hardware Accelerators
			9.3.1 Use of GPUs
			9.3.2 Use of FPGAs
			9.3.3 Dedicated AI Accelerators Using ASICs
		9.4 Cerebras´s Wafer Scale AI Engine
		9.5 Google Cloud TPUs
		9.6 Amazon´s Inference Engine
		9.7 Intel´s Movidius VPU
		9.8 Apple´s AI Ecosystem
		9.9 Summary
		9.10 Points to Ponder
		References
Appendix A
	AI/ML for App Store Predictions
		Using Python for App Metrics Predictions in Google Play Store
			Step 1: Obtain Data
			Step 2: Scrub and Prepare the Data
			Step 2.1: Null Value Removal/Fills
			Step 2.2: Duplicate Removal
			Exact Duplicates (876)
			Slight Variations (294)
			Step 2.3: Data Type Conversions
			Step 2.4: Dummy Variables
			Step 2.5: Rescale
			Step 3: Data Exploration
			Step 4: Categorization
			Step 4.1: Google App Ratings
			Step 4.2: Google App Reviews
			Step 4.3: Google App Sizes
			Step 4. 4: Google App Installs
			Step 4.5: Type
			Step 4.6: Price
			Step 4.7: Content Rating
			Step 4.8: Genres
			Step 4.9: Last Updated
			Step 4.10: Current Ver
			Step 4.11: Android Ver
			Step 4.12: Translated Review
			Step 4.13: Sentiment
			Step 4.14: Sentiment Polarity
			Step 4.15: Sentiment Subjectivity
			Step 4.16: Merging the Data
			Step 5: Results
			Step 5.1: Predicting Installs
			Step 5.2: Predict Category
			Step 5.3: Predict Rating
		Conclusion
	Migrating Python AI/ML Code to a Public Cloud
		Setting Up AWS Cloud Machine and Running a Python File on It
Appendix B: AI/ML for Letter Recognition
	Using Python for Letter Recognition
		Step 1: Obtain Data
		Step 2: Data Exploration and Attributes
		Step 3: Data Preprocessing
		Step 4: Data Classification Models
		Step 5: Feature Importance and Confusion Matrix
		Conclusion
		Future Work and Other Applications
	Migrating Code to Google Cloud Platform (GCP)
Appendix C: Supervised Learning for MNIST Digits Dataset
	Using Python ML Libraries for Letter Recognition
Appendix D: Points to Ponder
	Chapter 1: Points to Ponder
	Chapter 2: Points to Ponder
	Chapter 3: Points to Ponder
	Chapter 4: Points to Ponder
	Chapter 5: Points to Ponder
	Chapter 6: Points to Ponder
	Chapter 7: Points to Ponder
	Chapter 8: Points to Ponder
	Chapter 9: Points to Ponder
Appendix E: Questions for Practice
Glossary
Index