Data Science: Concepts and Practice

Cover
Data Science:
Concepts and Practice
Copyright
Dedication
Foreword
Preface
	Why Data Science?
	Why This Book?
	Who Can Use This Book?
Acknowledgments
1 Introduction
	1.1 AI, Machine learning, and Data Science
	1.2 What is Data Science?
		1.2.1 Extracting Meaningful Patterns
		1.2.2 Building Representative Models
		1.2.3 Combination of Statistics, Machine Learning, and Computing
		1.2.4 Learning Algorithms
		1.2.5 Associated Fields
	1.3 Case for Data Science
		1.3.1 Volume
		1.3.2 Dimensions
		1.3.3 Complex Questions
	1.4 Data Science Classification
	1.5 Data Science Algorithms
	1.6 Roadmap for This Book
		1.6.1 Getting Started With Data Science
		1.6.2 Practice using RapidMiner
		1.6.3 Core Algorithms
	References
2 Data Science Process
	2.1 Prior Knowledge
		2.1.1 Objective
		2.1.2 Subject Area
		2.1.3 Data
		2.1.4 Causation Versus Correlation
	2.2 Data Preparation
		2.2.1 Data Exploration
		2.2.2 Data Quality
		2.2.3 Missing Values
		2.2.4 Data Types and Conversion
		2.2.5 Transformation
		2.2.6 Outliers
		2.2.7 Feature Selection
		2.2.8 Data Sampling
	2.3 Modeling
		2.3.1 Training and Testing Datasets
		2.3.2 Learning Algorithms
		2.3.3 Evaluation of the Model
		2.3.4 Ensemble Modeling
	2.4 Application
		2.4.1 Production Readiness
		2.4.2 Technical Integration
		2.4.3 Response Time
		2.4.4 Model Refresh
		2.4.5 Assimilation
	2.5 Knowledge
	References
3 Data Exploration
	3.1 Objectives of Data Exploration
	3.2 Datasets
		3.2.1 Types of Data
			Numeric or Continuous
			Categorical or Nominal
	3.3 Descriptive Statistics
		3.3.1 Univariate Exploration
			Measure of Central Tendency
			Measure of Spread
		3.3.2 Multivariate Exploration
			Central Data Point
			Correlation
	3.4 Data Visualization
		3.4.1 Univariate Visualization
			Histogram
			Quartile
			Distribution Chart
		3.4.2 Multivariate Visualization
			Scatterplot
			Scatter Multiple
			Scatter Matrix
			Bubble Chart
			Density Chart
		3.4.3 Visualizing High-Dimensional Data
			Parallel Chart
			Deviation Chart
			Andrews Curves
	3.5 Roadmap for Data Exploration
	References
4 Classification
	4.1 Decision Trees
		4.1.1 How It Works
			Step 1: Where to Split Data?
			Step 2: When to Stop Splitting Data?
		4.1.2 How to Implement
			Implementation 1: To Play Golf or Not?
			Implementation 2: Prospect Filtering
				Step 1: Data Preparation
				Step 2: Divide dataset Into Training and Testing Samples
				Step 3: Modeling Operator and Parameters
				Step 4: Configuring the Decision Tree Model
				Step 5: Process Execution and Interpretation
		4.1.3 Conclusion
	4.2 Rule Induction
		WARNING!!! DUMMY ENTRY
			Approaches to Developing a Rule Set
		4.2.1 How It Works
			Step 1: Class Selection
			Step 2: Rule Development
			Step 3: Learn-One-Rule
			Step 4: Next Rule
			Step 5: Development of Rule Set
		4.2.2 How to Implement
			Step 1: Data Preparation
			Step 2: Modeling Operator and Parameters
			Step 3: Results Interpretation
			Alternative Approach: Tree-to-Rules
		4.2.3 Conclusion
	4.3 k-Nearest Neighbors
		4.3.1 How It Works
			Measure of Proximity
				Distance
				Weights
				Correlation similarity
				Simple matching coefficient
				Jaccard similarity
				Cosine similarity
		4.3.2 How to Implement
			Step 1: Data Preparation
			Step 2: Modeling Operator and Parameters
			Step 3: Execution and Interpretation
		4.3.3 Conclusion
	4.4 Naïve Bayesian
		4.4.1 How It Works
			Step 1: Calculating Prior Probability P(Y)
			Step 2: Calculating Class Conditional Probability P(Xi|Y)
			Step 3: Predicting the Outcome Using Bayes’ Theorem
			Issue 1: Incomplete Training Set
			Issue 2: Continuous Attributes
			Issue 3: Attribute Independence
		4.4.2 How to Implement
			Step 1: Data Preparation
			Step 2: Modeling Operator and Parameters
			Step 3: Evaluation
			Step 4: Execution and Interpretation
		4.4.3 Conclusion
	4.5 Artificial Neural Networks
		4.5.1 How It Works
			Step 1: Determine the Topology and Activation Function
			Step 2: Initiation
			Step 3: Calculating Error
			Step 4: Weight Adjustment
		4.5.2 How to Implement
			Step 1: Data Preparation
			Step 2: Modeling Operator and Parameters
			Step 3: Evaluation
			Step 4: Execution and Interpretation
		4.5.3 Conclusion
	4.6 Support Vector Machines
		WARNING!!! DUMMY ENTRY
			Concept and Terminology
		4.6.1 How It Works
		4.6.2 How to Implement
			Implementation 1: Linearly Separable Dataset
				Step 1: Data Preparation
				Step 2: Modeling Operator and Parameters
				Step 3: Process Execution and Interpretation
			Example 2: Linearly Non-Separable Dataset
				Step 1: Data Preparation
				Step 2: Modeling Operator and Parameters
				Step 3: Execution and Interpretation
			Parameter Settings
		4.6.3 Conclusion
	4.7 Ensemble Learners
		WARNING!!! DUMMY ENTRY
			Wisdom of the Crowd
		4.7.1 How It Works
			Achieving the Conditions for Ensemble Modeling
		4.7.2 How to Implement
			Ensemble by Voting
			Bootstrap Aggregating or Bagging
			Implementation
			Boosting
			AdaBoost
			Implementation
			Random Forest
			Implementation
		4.7.3 Conclusion
	References
5 Regression Methods
	5.1 Linear Regression
		5.1.1 How it Works
		5.1.2 How to Implement
			Step 1: Data Preparation
			Step 2: Model Building
			Step 3: Execution and Interpretation
			Step 4: Application to Unseen Test Data
		5.1.3 Checkpoints
	5.2 Logistic Regression
		5.2.1 How It Works
			How Does Logistic Regression Find the Sigmoid Curve?
			A Simple but Tragic Example
		5.2.2 How to Implement
			Step 1: Data Preparation
			Step 2: Modeling Operator and Parameters
			Step 3: Execution and Interpretation
			Step 4: Using MetaCost
			Step 5: Applying the Model to an Unseen Dataset
		5.2.3 Summary Points
	5.3 Conclusion
	References
6 Association Analysis
	6.1 Mining Association Rules
		6.1.1 Itemsets
			Support
			Confidence
			Lift
			Conviction
		6.1.2 Rule Generation
	6.2 Apriori Algorithm
		6.2.1 How it Works
			Frequent Itemset Generation
			Rule Generation
	6.3 Frequent Pattern-Growth Algorithm
		6.3.1 How it Works
			Frequent Itemset Generation
		6.3.2 How to Implement
			Step 1: Data Preparation
			Step 2: Modeling Operator and Parameters
			Step 3: Create Association Rules
			Step 4: Interpreting the Results
	6.4 Conclusion
	References
7 Clustering
	Clustering to Describe the Data
	Clustering for Preprocessing
	Types of Clustering Techniques
	7.1 k-Means Clustering
		7.1.1 How It Works
			Step 1: Initiate Centroids
			Step 2: Assign Data Points
			Step 3: Calculate New Centroids
			Step 4: Repeat Assignment and Calculate New Centroids
			Step 5: Termination
			Special Cases
			Evaluation of Clusters
		7.1.2 How to Implement
			Step 1: Data Preparation
			Step 2: Clustering Operator and Parameters
			Step 3: Evaluation
			Step 4: Execution and Interpretation
	7.2 DBSCAN Clustering
		7.2.1 How It Works
			Step 1: Defining Epsilon and MinPoints
			Step 2: Classification of Data Points
			Step 3: Clustering
			Optimizing Parameters
			Special Cases: Varying Densities
		7.2.2 How to Implement
			Step 1: Data Preparation
			Step 2: Clustering Operator and Parameters
			Step 3: Evaluation
			Step 4: Execution and Interpretation
	7.3 Self-Organizing Maps
		7.3.1 How It Works
			Step 1: Topology Specification
			Step 2: Initialize Centroids
			Step 3: Assignment of Data Objects
			Step 4: Centroid Update
			Step 5: Termination
			Step 6: Mapping a New Data Object
		7.3.2 How to Implement
			Step 1: Data Preparation
			Step 2: SOM Modeling Operator and Parameters
			Step 3: Execution and Interpretation
			Visual Model
			Location Coordinates
			Conclusion
	References
8 Model Evaluation
	8.1 Confusion Matrix
	8.2 ROC and AUC
	8.3 Lift Curves
	8.4 How to Implement
		WARNING!!! DUMMY ENTRY
			Step 1: Data Preparation
			Step 2: Modeling Operator and Parameters
			Step 3: Evaluation
			Step 4: Execution and Interpretation
	8.5 Conclusion
	References
9 Text Mining
	9.1 How It Works
		9.1.1 Term Frequency–Inverse Document Frequency
		9.1.2 Terminology
	9.2 How to Implement
		9.2.1 Implementation 1: Keyword Clustering
			Step 1: Gather Unstructured Data
			Step 2: Data Preparation
			Step 3: Apply Clustering
		9.2.2 Implementation 2: Predicting the Gender of Blog Authors
			Step 1: Gather Unstructured Data
			Step 2: Data Preparation
			Step 3.1: Identify Key Features
			Step 3.2: Build Models
			Step 4.1: Prepare Test Data for Model Application
			Step 4.2: Applying the Trained Models to Testing Data
			Bias in Machine Learning
	9.3 Conclusion
	References
10 Deep Learning
	10.1 The AI Winter
		AI Winter: 1970’s
		Mid-Winter Thaw of the 1980s
		The Spring and Summer of Artificial Intelligence: 2006—Today
	10.2 How it works
		10.2.1 Regression Models As Neural Networks
		10.2.2 Gradient Descent
		10.2.3 Need for Backpropagation
		10.2.4 Classifying More Than 2 Classes: Softmax
		10.2.5 Convolutional Neural Networks
		10.2.6 Dense Layer
		10.2.7 Dropout Layer
		10.2.8 Recurrent Neural Networks
		10.2.9 Autoencoders
		10.2.10 Related AI Models
	10.3 How to Implement
		WARNING!!! DUMMY ENTRY
			Handwritten Image Recognition
			Step 1: Dataset Preparation
			Step 2: Modeling using the Keras Model
			Step 3: Applying the Keras Model
			Step 4: Results
	10.4 Conclusion
	References
11 Recommendation Engines
	Why Do We Need Recommendation Engines?
	Applications of Recommendation Engines
	11.1 Recommendation Engine Concepts
		WARNING!!! DUMMY ENTRY
			Building up the Ratings Matrix
			Step 1: Assemble Known Ratings
			Step 2: Rating Prediction
			Step 3: Evaluation
			The Balance
		11.1.1 Types of Recommendation Engines
	11.2 Collaborative Filtering
		11.2.1 Neighborhood-Based Methods
			User-Based Collaborative Filtering
			Step 1: Identifying Similar Users
			Step 2: Deducing Rating From Neighborhood Users
			Item-Based Collaborative Filtering
			User-Based or Item-Based Collaborative Filtering?
			Neighborhood based Collaborative Filtering - How to Implement
			Dataset
			Implementation Steps
			Conclusion
		11.2.2 Matrix Factorization
			Matrix Factorization - How to Implement
			Implementation Steps
	11.3 Content-Based Filtering
		WARNING!!! DUMMY ENTRY
			Building an Item Profile
		11.3.1 User Profile Computation
			Content-Based Filtering - How to Implement
			Dataset
			Implementation steps
		11.3.2 Supervised Learning Models
			Supervised Learning Models - How to Implement
			Dataset
			Implementation steps
	11.4 Hybrid Recommenders
	11.5 Conclusion
		WARNING!!! DUMMY ENTRY
			Summary of the Types of Recommendation Engines
	References
12 Time Series Forecasting
	Taxonomy of Time Series Forecasting
	12.1 Time Series Decomposition
		12.1.1 Classical Decomposition
		12.1.2 How to Implement
			Forecasting Using Decomposed Data
	12.2 Smoothing Based Methods
		12.2.1 Simple Forecasting Methods
			Naïve Method
			Seasonal Naive Method
			Average Method
			Moving Average Smoothing
			Weighted Moving Average Smoothing
		12.2.2 Exponential Smoothing
			Holt’s Two-Parameter Exponential Smoothing
			Holt-Winters’ Three-Parameter Exponential Smoothing
		12.2.3 How to Implement
			R Script for Holt-Winters’ Forecasting
	12.3 Regression Based Methods
		12.3.1 Regression
		12.3.2 Regression With Seasonality
			How to implement
		12.3.3 Autoregressive Integrated Moving Average
			Autocorrelation
			Autoregressive Models
			Stationary Data
			Differencing
			Moving Average of Error
			Autoregressive Integrated Moving Average
			How to Implement
		12.3.4 Seasonal ARIMA
			How to Implement
	12.4 Machine Learning Methods
		12.4.1 Windowing
			Model Training
			How to Implement
			Step 1: Set Up Windowing
			Step 2: Train the Model
			Step 3: Generate the Forecast in a Loop
		12.4.2 Neural Network Autoregressive
			How to Implement
	12.5 Performance Evaluation
		12.5.1 Validation Dataset
			Mean Absolute Error
			Root Mean Squared Error
			Mean Absolute Percentage Error
			Mean Absolute Scaled Error
		12.5.2 Sliding Window Validation
	12.6 Conclusion
		12.6.1 Forecasting Best Practices
	References
13 Anomaly Detection
	13.1 Concepts
		13.1.1 Causes of Outliers
		13.1.2 Anomaly Detection Techniques
			Outlier Detection Using Statistical Methods
			Outlier Detection Using Data Science
	13.2 Distance-Based Outlier Detection
		13.2.1 How It Works
		13.2.2 How to Implement
			Step 1: Data Preparation
			Step 2: Detect Outlier Operator
			Step 3: Execution and Interpretation
	13.3 Density-Based Outlier Detection
		13.3.1 How It Works
		13.3.2 How to Implement
			Step 1: Data Preparation
			Step 2: Detect Outlier Operator
			Step 3: Execution and Interpretation
	13.4 Local Outlier Factor
		13.4.1 How it Works
		13.4.2 How to Implement
			Step 1: Data Preparation
			Step 2: Detect Outlier Operator
			Step 3: Results Interpretation
	13.5 Conclusion
	References
14 Feature Selection
	14.1 Classifying Feature Selection Methods
	14.2 Principal Component Analysis
		14.2.1 How It Works
		14.2.2 How to Implement
			Step 1: Data Preparation
			Step 2: PCA Operator
			Step 3: Execution and Interpretation
	14.3 Information Theory-Based Filtering
	14.4 Chi-Square-Based Filtering
	14.5 Wrapper-Type Feature Selection
		14.5.1 Backward Elimination
	14.6 Conclusion
	References
15 Getting Started with RapidMiner
	15.1 User Interface and Terminology
		WARNING!!! DUMMY ENTRY
			Terminology
	15.2 Data Importing and Exporting Tools
	15.3 Data Visualization Tools
		WARNING!!! DUMMY ENTRY
			Univariate Plots
			Bivariate Plots
			Multivariate Plots
	15.4 Data Transformation Tools
	15.5 Sampling and Missing Value Tools
	15.6 Optimization Tools5
	15.7 Integration with R
	15.8 Conclusion
	References
Comparison of Data Science Algorithms
About the Authors
	Vijay Kotu
	Bala Deshpande, PhD
Index
Praise
Back Cover