reading

Cover
Preface
Part I: Introductory Background
1. 1 What Can We Do With Data?
  1. 1.1 Big Data and Data Science
  2. 1.2 Big Data Architectures
  3. 1.3 Small Data
  4. 1.4 What is Data?
  5. 1.5 A Short Taxonomy of Data Analytics
  6. 1.6 Examples of Data Use
  7. 1.7 A Project on Data Analytics
  8. 1.8 How this Book is Organized
  9. 1.9 Who Should Read this Book
Part II: Getting Insights from Data
1. 2 Descriptive Statistics
  1. 2.1 Scale Types
  2. 2.2 Descriptive Univariate Analysis
  3. 2.3 Descriptive Bivariate Analysis
  4. 2.4 Final Remarks
  5. 2.5 Exercises
2. 3 Descriptive Multivariate Analysis
  1. 3.1 Multivariate Frequencies
  2. 3.2 Multivariate Data Visualization
  3. 3.3 Multivariate Statistics
  4. 3.4 Infographics and Word Clouds
  5. 3.5 Final Remarks
  6. 3.6 Exercises
3. 4 Data Quality and Preprocessing
  1. 4.1 Data Quality
  2. 4.2 Converting to a Different Scale Type
  3. 4.3 Converting to a Different Scale
  4. 4.4 Data Transformation
  5. 4.5 Dimensionality Reduction
  6. 4.6 Final Remarks
  7. 4.7 Exercises
4. 5 Clustering
  1. 5.1 Distance Measures
  2. 5.2 Clustering Validation
  3. 5.3 Clustering Techniques
  4. 5.4 Final Remarks
  5. 5.5 Exercises
5. 6 Frequent Pattern Mining
  1. 6.1 Frequent Itemsets
  2. 6.2 Association Rules
  3. 6.3 Behind Support and Confidence
  4. 6.4 Other Types of Pattern
  5. 6.5 Final Remarks
  6. 6.6 Exercises
6. 7 Cheat Sheet and Project on Descriptive Analytics
  1. 7.1 Cheat Sheet of Descriptive Analytics
  2. 7.2 Project on Descriptive Analytics
Part III: Predicting the Unknown
1. 8 Regression
  1. 8.1 Predictive Performance Estimation
  2. 8.2 Finding the Parameters of the Model
  3. 8.3 Technique and Model Selection
  4. 8.4 Final Remarks
  5. 8.5 Exercises
2. 9 Classification
  1. 9.1 Binary Classification
  2. 9.2 Predictive Performance Measures for Classification
  3. 9.3 Distance‐based Learning Algorithms
  4. 9.4 Probabilistic Classification Algorithms
  5. 9.5 Final Remarks
  6. 9.6 Exercises
3. 10 Additional Predictive Methods
  1. 10.1 Search‐based Algorithms
  2. 10.2 Optimization‐based Algorithms
  3. 10.3 Final Remarks
  4. 10.4 Exercises
4. 11 Advanced Predictive Topics
  1. 11.1 Ensemble Learning
  2. 11.2 Algorithm Bias
  3. 11.3 Non‐binary Classification Tasks
  4. 11.4 Advanced Data Preparation Techniques for Prediction
  5. 11.5 Description and Prediction with Supervised Interpretable Techniques
  6. 11.6 Exercises
5. 12 Cheat Sheet and Project on Predictive Analytics
  1. 12.1 Cheat Sheet on Predictive Analytics
  2. 12.2 Project on Predictive Analytics
Part IV: Popular Data Analytics Applications
1. 13 Applications for Text, Web and Social Media
  1. 13.1 Working with Texts
  2. 13.2 Recommender Systems
  3. 13.3 Social Network Analysis
  4. 13.4 Exercises
Appendix A: A Comprehensive Description of the CRISP‐DM Methodology
1. A.1 Business Understanding
2. A.2 Data Understanding
3. A.3 Data Preparation
4. A.4 Modeling
5. A.5 Evaluation
6. A.6 Deployment
References
Index
End User License Agreement

List of Tables

Chapter 01
1. Table 1.1 Data set of our private contact list.
2. Table 1.2 Family relations between contacts.
Chapter 02
1. Table 2.1 Data set of our private list of contacts with weight and height.
2. Table 2.2 Univariate absolute and relative frequencies for “company” attribute.
3. Table 2.3 Univariate absolute and relative frequencies for height.
4. Table 2.4 Univariate plots.
5. Table 2.5 Location univariate statistics for weight.
6. Table 2.6 Central tendency statistics according to the type of scale.
7. Table 2.7 Dispersion univariate statistics for the “weight” attribute.
8. Table 2.8 The rank values for the attributes “weight” and “height”.
Chapter 03
1. Table 3.1 Data set of our private list of contacts with weight and height.
2. Table 3.2 Location univariate statistics for quantitative attributes.
3. Table 3.3 Dispersion univariate statistics for quantitative attributes.
4. Table 3.4 Covariance matrix for quantitative attributes.
5. Table 3.5 Pearson correlation matrix for quantitative attributes.
Chapter 04
1. Table 4.1 Filling of missing values.
2. Table 4.2 Removal of redundant objects.
3. Table 4.3 Data set of our private list of contacts with weight and height.
4. Table 4.4 Food preferences of our colleagues.
5. Table 4.5 Conversion from nominal scale to relative scale.
6. Table 4.6 Conversion from the nominal scale to binary values.
7. Table 4.7 Conversion from the nominal scale to the relative scale.
8. Table 4.8 Conversion from the nominal scale to the relative scale.
9. Table 4.9 Conversion from the nominal scale to the relative scale.
10. Table 4.10 Conversion from the ordinal scale to the relative or absolute scale.
11. Table 4.11 Conversion from the ordinal scale to the relative scale.
12. Table 4.12 Euclidean distances of ages expressed in years.
13. Table 4.13 Euclidean distance with age expressed in decades.
14. Table 4.14 Normalization using min–max rescaling.
15. Table 4.15 Normalization using standardization.
16. Table 4.16 Euclidean distance with normalized values.
17. Table 4.17 How much each friend earns as income and spends on dinners per year.
18. Table 4.18 Correlation between each predictive attribute and the target attribute.
19. Table 4.19 Predictive performance of a classifier for each predictive attribute.
20. Table 4.20 Filling of missing values.
Chapter 05
1. Table 5.1 Simple social network data set.
2. Table 5.2 Example of bag of words vectors.
3. Table 5.3 The clusters to which each friend belongs to according to Figure 5.7a–f.
4. Table 5.4 Normalized centroids of the example dataset.
5. Table 5.5 Advantages and disadvantages of k‐means.
6. Table 5.6 Advantages and disadvantages of DBSCAN.
7. Table 5.7 First iteration of agglomerative hierarchical clustering.
8. Table 5.8 Second iteration of agglomerative hierarchical clustering.
9. Table 5.9 Advantages and disadvantages of agglomerative hierarchical clustering.
Chapter 06
1. Table 6.1 Data about the preferred cuisine of contacts.
2. Table 6.2 Transactional data created from Table 6.1.
3. Table 6.3 Combinatorial explosion with growing size of .
4. Table 6.4 Transactional data from Table 6.2 in vertical format corresponding to the first iteration () of the Eclat algorithm.
5. Table 6.5 An example of a two‐way contingency table for itemsets and .
6. Table 6.6 Two‐way contingency tables for itemsets and for two groups A and B of data and the whole data set (combined groups A and B). The presence or absence of itemsets in transactions are marked by Yes and No, respectively.
7. Table 6.7 Example of a sequence database with items .
Chapter 07
1. Table 7.1 Summarizing methods for a single attribute.
2. Table 7.2 Summarizing methods for two attributes.
3. Table 7.3 Distance measures.
4. Table 7.4 Clustering methods.
5. Table 7.5 Time complexity and memory requirements of frequent itemset mining approaches.
6. Table 7.6 Measures, generally related to frequent mining approaches.
7. Table 7.7 Attributes of the Breast Cancer Wisconsin data set.
8. Table 7.8 Statistics of the attributes of the Breast Cancer Wisconsin data set.
Chapter 08
1. Table 8.1 Data set of our contact list, with weights and heights.
2. Table 8.2 Advantages and disadvantages of linear regression.
3. Table 8.3 Advantages and disadvantages of ridge regression.
4. Table 8.4 Advantages and disadvantages of the lasso.
5. Table 8.5 My social network data using the height as target.
Chapter 09
1. Table 9.1 Simple labeled social network data set for model induction.
2. Table 9.2 Simple labeled social network data set 2.
3. Table 9.3 Simple labeled social network data set 3.
4. Table 9.4 Extended simple labeled social network data set.
5. Table 9.5 Advantages and disadvantages of k‐NN.
6. Table 9.6 Advantages and disadvantages of logistic regression.
7. Table 9.7 Advantages and disadvantages of NB
Chapter 10
1. Table 10.1 Simple labeled social network data set 3.
2. Table 10.2 Advantages and disadvantages of decision trees.
3. Table 10.3 Advantages and disadvantages of MARS.
4. Table 10.4 Advantages and disadvantages of ANNs.
5. Table 10.5 Advantages and disadvantages of DL.
6. Table 10.6 Advantages and disadvantages of SVM.
Chapter 11
1. Table 11.1 Advantages and disadvantages of bagging.
2. Table 11.2 Advantages and disadvantages of random forests.
3. Table 11.3 Advantages and disadvantages of Adaboost.
Chapter 12
1. Table 12.1 A cheat sheet on predictive algorithms.
2. Table 12.2 Predictive attributes of the Polish company insolvency data set.
3. Table 12.3 Statistics of the Polish company insolvency data set.
4. Table 12.4 K‐NN confusion matrix using five predictive attributes.
5. Table 12.5 C4.5 confusion matrix using five predictive attributes.
6. Table 12.6 Random forest confusion matrix using all predictive attributes.
Chapter 13
1. Table 13.1 Training set of labeled texts.
2. Table 13.2 Results of stemming.
3. Table 13.3 Results of applying a stemmer.
4. Table 13.4 Stems after removal of stop words.
5. Table 13.5 Objects with their stems.
6. Table 13.6 Item recommendation scenario.
7. Table 13.7 Rating prediction scenario.
8. Table 13.8 Data for a content‐based RT for the user Eve from Table 13.7.
9. Table 13.9 Cosine vector similarities between the users from Table 13.6.
10. Table 13.10 Pearson correlation similarities between users in Table 13.7.
11. Table 13.11 The adjacency matrix for the network in Figure 13.5.
12. Table 13.12 The adjacency matrix from the Table 13.11 squared showing the counts of paths of length two between pairs of nodes.
13. Table 13.13 Basic properties of nodes from the network in Figure 13.5.
14. Table 13.14 The distance matrix – distances between nodes – for the graph in Figure 13.5.

List of Illustrations

Chapter 01
1. Figure 1.1 A prediction model to classify someone as either good or bad company.
2. Figure 1.2 The use of different methodologies on data analytics through time.
3. Figure 1.3 The CRISP‐DM methodology
Chapter 02
1. Figure 2.1 The main areas of statistics.
2. Figure 2.2 The relation between the four types of scales: absolute, relative, ordinal and nominal.
3. Figure 2.3 An example of an area chart used to compare several probability density functions.
4. Figure 2.4 Price absolute frequency distributions with (histogram) and without (bar chart) cell definition.
5. Figure 2.5 Empirical and probability distribution functions.
6. Figure 2.6 Stacked bar plot for “company” split by “gender”.
7. Figure 2.7 Location statistics on the absolute frequency plot for the attribute “weight”.
8. Figure 2.8 Box‐plot for the attribute “height”.
9. Figure 2.9 Central tendency statistics in asymmetric and symmetric unimodal distributions.
10. Figure 2.10 An example of the Likert scale.
11. Figure 2.11 Combination of a histogram and a box‐plot for the “height” attribute.
12. Figure 2.12 The probability density function, of
13. Figure 2.13 The probability density function for different standard deviations, .
14. Figure 2.14 3D histogram for attributes “weight” and “height”.
15. Figure 2.15 Scatter plot for the attributes “weight” and “height”.
16. Figure 2.16 Three examples of correlation between two attributes.
17. Figure 2.17 The scatter plot for the attributes and .
18. Figure 2.18 Contingency table with absolute joint frequencies for “company” and “gender”.
19. Figure 2.19 Mosaic plot for “company” and “gender”.
20. Figure 2.20 Scatter plot.
Chapter 03
1. Figure 3.1 Plot of objects with three attributes.
2. Figure 3.2 Two alternatives for a plot of three attributes, where the the third attribute is qualitative.
3. Figure 3.3 Plot for three attributes from the contacts data set.
4. Figure 3.4 Plot for four attributes of the friends data set using color for the forth attribute.
5. Figure 3.5 Parallel coordinate plot for three attributes.
6. Figure 3.6 Parallel coordinate plot for five attributes.
7. Figure 3.7 Parallel coordinate plots for multiple attributes: left, using a different style of line for contacts who are good and bad company; right, with the order of the attributes changed as well.
8. Figure 3.8 Star plot with the value of each attribute for each object in our contacts data set.
9. Figure 3.9 Star plot with the value of each attribute for each object in contacts data set.
10. Figure 3.10 Visualization of the objects in our contacts data set using Chernoff faces.
11. Figure 3.11 Set of box plots, one for each attribute.
12. Figure 3.12 Matrix of scatter plots for quantitative attributes.
13. Figure 3.13 Matrix of scatter plots for quantitative attributes with additional Pearson correlation values.
14. Figure 3.14 Correlograms for Pearson correlation between the attributes “maxtemp”, “weight”, “height” and “years”.
15. Figure 3.15 Heatmap for the short version of the contacts data set.
16. Figure 3.16 Infographic of the level of qualifications in England (Contains public sector information licensed under the Open Government Licence v3.0.).
17. Figure 3.17 Text visualization using a word cloud.
Chapter 04
1. Figure 4.1 Data set with and without noise.
2. Figure 4.2 Data set with outliers.
3. Figure 4.3 Outlier detection based on the interquartile range distance.
4. Figure 4.4 Two alternatives for a plot for three attributes the last of which is qualitative.
5. Figure 4.5 Principal components obtained by PCA for the short version of the contacts data set.
6. Figure 4.6 Components obtained by PCA and ICA for the short version of the contacts data set.
Chapter 05
1. Figure 5.1 Applying clustering to a dataset.
2. Figure 5.2 Alternative clustering of the data set.
3. Figure 5.3 Euclidean and Manhattan distances.
4. Figure 5.4 Dynamic time warping.
5. Figure 5.5 Transformation of images of size pixels into matrices and vectors.
6. Figure 5.6 (a) Partitional and (b) hierarchical clustering.
7. Figure 5.7 K‐means with . The big symbols represent the centroids. The example is step‐by‐step according to the Algorithm K‐means: (a) step 4; (b) 1st iteration of step 6; (c) 1st iteration of step 7; (d) 2nd iteration of step 6; (e) 2nd iteration of step 7; and (f) 3rd iteration of step (6). The algorithm stops in the 3rd iteration (f) because there is no instance changing of symbol between the 2nd (d) and the 3rd (f) iterations of step (6).
8. Figure 5.8 Examples of convex and non‐convex shapes.
9. Figure 5.9 Variation of within‐groups sum of squares with number of clusters.
10. Figure 5.10 DBSCAN for the example data set using a minimum number of instances of two.
11. Figure 5.11 Comparison of k‐means and DBSCAN: (a) k‐means with ; (b) DBSCAN with minimum number of instances of and epsilon of .
12. Figure 5.12 Single, complete and average linkage criteria.
13. Figure 5.13 Effect of different linkage criteria on sample data set: (a) single; (b) complete; (c) average; (d) Ward.
14. Figure 5.14 Dendrogram defined by agglomerative hierarchical clustering using the average as linkage criterion.
15. Figure 5.15 Dendrogram.
16. Figure 5.16 Space invaders.
Chapter 06
1. Figure 6.1 Itemsets organized into a lattice according to subset‐superset relations. Members of are abbreviations for the five cuisine types introduced in Table 6.2, i.e. A for the Arabic cuisine, I for Indian, etc.
2. Figure 6.2 The Apriori principle on the data from Table 6.2 (or 6.1), illustrated using an enumeration tree.
3. Figure 6.3 Building an FP‐tree on the data from Table 6.2 (or 6.1). Each node contains an identifier of an item and its count in the given path.
4. Figure 6.4 The FP‐growth process of finding frequent itemsets ending in item O but not containing item A from the FT‐Tree (Figure 6.3) built from the data in Table 6.2.
5. Figure 6.5 Maximal (shaded) and closed (with solid edges) itemsets generated from the data in Table 6.2.
6. Figure 6.6 Association rules lattice corresponding to the frequent itemset {I,M,O} found in the data (Table 6.2).
Chapter 07
1. Figure 7.1 Relationship between frequent, closed and maximal patterns.
2. Figure 7.2 Centroids for k‐means with , where the benign breast mass is Cluster 1.
3. Figure 7.3 Centroids for k‐means with , where the benign breast mass is Cluster 1.
4. Figure 7.4 Centroids for k‐means with , where the benign breast mass is Clusters 1 and 2.
5. Figure 7.5 The elbow curve for the breast cancer problem using the average of the within‐groups sum of squares.
Chapter 08
1. Figure 8.1 Simple linear regression for the social network data.
2. Figure 8.2 Holdout validation method.
3. Figure 8.3 Random sub‐sampling validation method.
4. Figure 8.4 k‐fold cross validation method.
5. Figure 8.5 Leave one out method.
6. Figure 8.6 Bootstrap method.
7. Figure 8.7 Approximately homogeneous variance between diameter and height and nonhomogeneous variance between diameter and age. This data is about the shellfish Haliotis moluscus, also known as abalone. It is a food source in several human cultures. The shells are used as decorative items.
8. Figure 8.8 Performance of different models when trained and tested on different instances of the same data set. Refer to main text for details.
9. Figure 8.9 Models from the Figure 8.8: left, average models; center, linear regression models; right, polynomial models.
10. Figure 8.10 The bias–variance trade‐off.
Chapter 09
1. Figure 9.1 Simple binary classification task.
2. Figure 9.2 Classification model induced for the binary classification task.
3. Figure 9.3 New binary classification task.
4. Figure 9.4 Binary classification task with two predictive attributes.
5. Figure 9.5 Classification model induced by a classification algorithm for the classification task with two predictive attributes.
6. Figure 9.6 Linearly separable classification task with three predictive attributes.
7. Figure 9.7 Confusion matrix for a binary classification task.
8. Figure 9.8 Data distribution according to predictive attribute values, true class and predicted class for a fictitious dataset.
9. Figure 9.9 Predictive performance measures for a binary classification task.
10. Figure 9.10 Confusion matrix for the example dataset.
11. Figure 9.11 ROC graph for three classifiers.
12. Figure 9.12 Classifier AUC with different ROC values.
13. Figure 9.13 Different AUC situations.
14. Figure 9.14 Influence of the value of in k‐NN algorithm classification.
15. Figure 9.15 Case‐based reasoning, adapted from [29].
16. Figure 9.16 The age empirical distributions for people who are good and bad company.
Chapter 10
1. Figure 10.1 Example of a decision tree.
2. Figure 10.2 Input space partition by a decision tree.
3. Figure 10.3 Graphical representation of the CART classification tree for our social data set in Table 8.5.
4. Figure 10.4 Rules representation of the CART classification tree for our social data set from Table 8.5.
5. Figure 10.5 Example: , where is the prediction of produced by a CART decision.
6. Figure 10.6 Solving an artificial data set with (a) MLR, (b) CART, (c) model trees and (d) MARS.
7. Figure 10.7 Detail of a model tree and MARS using an artificial data set.
8. Figure 10.8 MLP neural network with two hidden layers.
9. Figure 10.9 How the MLP training affects the orientation and position of separating hyperplanes and the orientation and position of convex and non‐convex separating regions.
10. Figure 10.10 Orientation and position of separating hyperplanes, and orientation and position of convex and non‐convex separating regions of a trained MLP network.
11. Figure 10.11 Separation regions created by (left) an MLP network trained with backpropagation and (right) a decision tree induction algorithm for the same data set.
12. Figure 10.12 Large margin defined by SVM support vectors.
13. Figure 10.13 Decision borders found by (left) perceptron and (right) SVM learning algorithms.
14. Figure 10.14 Increase of separation margin by allowing some objects to be placed inside the separation margin.
15. Figure 10.15 Example of the use of a kernel function to transform a non‐linearly separable classification task in a two‐dimensional space to a linearly separable classification task in a three‐dimensional space.
16. Figure 10.16 The soft‐margin in a regression problem and how is calculated.
Chapter 11
1. Figure 11.1 Ensemble with a parallel combination of classifiers.
2. Figure 11.2 Ensemble with a sequential combination of classifiers.
3. Figure 11.3 Ensemble combining parallel and sequential approaches.
4. Figure 11.4 Example of a random forest predictive model.
5. Figure 11.5 Representational bias for three different classification algorithms.
6. Figure 11.6 Classification model induced by a decision tree algoritm on a data set with two predictive attributes.
7. Figure 11.7 Example of a one‐class classification task.
8. Figure 11.8 Example of a multi‐class classification task.
9. Figure 11.9 Example of a ranking classification task.
10. Figure 11.10 Example of a multi‐label classification task.
11. Figure 11.11 Example of a hierarchical classification task.
Chapter 13
1. Figure 13.1 Text about the food preferences of a friend.
2. Figure 13.2 Regression tree model learned from the training data in Table 13.8.
3. Figure 13.3 Representation of users and items in a common, latent two‐dimensional space from Example 13.9.
4. Figure 13.4 Undirected (a), weighted (b) where the thickness of an edge is proportional to its weight, directed (c), and, directed and weighted (d) edge types in networks.
5. Figure 13.5 An example social network. The nodes correspond to our friends from the example in Table 1.1; each person is indicated by their initial.
6. Figure 13.6 Part of a social network from Figure 13.5.
7. Figure 13.7 Example with three networks.

This edition first published 2019
© 2019 John Wiley & Sons, Inc.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by law. Advice on how to obtain permission to reuse material from this title is available at http://www.wiley.com/go/permissions.

The right of João Moreira, André de Carvalho, and Tomáš Horváth to be identified as the author(s) of this work has been asserted in accordance with law.

Registered Office
John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA

Editorial Office
111 River Street, Hoboken, NJ 07030, USA

For details of our global editorial offices, customer services, and more information about Wiley products visit us at www.wiley.com.

Wiley also publishes its books in a variety of electronic formats and by print-on-demand. Some content that appears in standard print versions of this book may not be available in other formats.

Limit of Liability/Disclaimer of Warranty
In view of ongoing research, equipment modifications, changes in governmental regulations, and the constant flow of information relating to the use of experimental reagents, equipment, and devices, the reader is urged to review and evaluate the information provided in the package insert or instructions for each chemical, piece of equipment, reagent, or device for, among other things, any changes in the instructions or indication of usage and for added warnings and precautions. While the publisher and authors have used their best efforts in preparing this work, they make no representations or warranties with respect to the accuracy or completeness of the contents of this work and specifically disclaim all warranties, including without limitation any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives, written sales materials or promotional statements for this work. The fact that an organization, website, or product is referred to in this work as a citation and/or potential source of further information does not mean that the publisher and authors endorse the information or services the organization, website, or product may provide or recommendations it may make. This work is sold with the understanding that the publisher is not engaged in rendering professional services. The advice and strategies contained herein may not be suitable for your situation. You should consult with a specialist where appropriate. Further, readers should be aware that websites listed in this work may have changed or disappeared between when this work was written and when it is read. Neither the publisher nor authors shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages.

Library of Congress Cataloging-in-Publication Data

Names: Moreira, João, 1969– author. | Carvalho, André Carlos Ponce de Leon Ferreira, author. | Horváth, Tomáš, 1976– author.
Title: A general introduction to data analytics / by João Mendes Moreira, André C. P. L. F. de Carvalho, Tomáš Horváth.
Description: Hoboken, NJ : John Wiley & Sons, 2019. | Includes bibliographical references and index. |
Identifiers: LCCN 2017060728 (print) | LCCN 2018005929 (ebook) | ISBN 9781119296256 (pdf) | ISBN 9781119296263 (epub) | ISBN 9781119296249 (cloth)
Subjects: LCSH: Mathematical statistics–Methodology. | Electronic data processing. | Data mining.
Classification: LCC QA276.4 (ebook) | LCC QA276.4 .M664 2018 (print) | DDC 519.50285–dc23
LC record available at https://lccn.loc.gov/2017060728

A General Introduction to Data Analytics

Preface

Acknowledgments

Presentational Conventions

About the Companion Website

Part I
Introductory Background