Discovering Knowledge in Data : An Introduction to Data Mining

There is a lot to like about this book, but it has some unfortunate flaws. Note that it is part of a Data Mining trilogy. The other two books are: Data Mining Methods and Models and Data Mining the Web: Uncovering Patterns in Web Content, Structure, and Usage. My initial reaction was more negative as I feel strongly about the issues that this book addresses poorly. However, I find myself turning to this book again and again. I would endorse it highly, but with a caution or two. The very best features of the book are the exceptionally clear explanations of complicated algorithms. In particular, Chapters 6 and 7 and their explanations of Decision Trees and Neural Nets are just perfect for both new and veteran analysts who want to understand what is happening "under the hood". Those chapters are stand-outs, but all of the 80%+ part of the book that describes algorithms in detail (clear, careful, and readable detail) is uniformly excellent. For some readers, it may be the first time that the techniques really make sense to them. Now the flaws. The three book format is, frankly, annoying. The second book and third books are much weaker, but the it was clearly designed as a trilogy, so it is hard to recommend the first to a client without at least implicitly recommending the second. Spending my reading time well is more important to me than my reading budget, but the set of three costs more than $200. Unless you plan on an entire shelf of related books, like me, I can't recommend the entire set. The other flaw is less obvious, and is the one that concerns me the most. Although this book cites Dorian Pyle's excellent book ... it seems to miss the whole point. Data Mining data prep is quite different from data prep for statistics. Although the two areas share a lot in common, and while mastery of statistics is a good thing for data miners, this is one of the differences between the two disciplines. Data cleaning and data reduction are critical, but this book suggests that this is accomplished by the human doing all possible bivariates. Recommendations of factor analysis and log transformations abound, but never with cautions of when that is unnecessary or even a bad idea - something Pyle's book explores. Also, transformations like binning come off as something the analyst does during data exploration, getting it perfect before modeling. Sounds like statistics data prep to me - not data mining data prep. If anyone has ever completed data prep without preliminary modeling, or has modeled without having to revisit data prep, I have never heard of it. If a novice data miner were to take the advice too literally, they could get themselves into trouble. This would be especially true of a reader that is well versed in statistics - there is a predictable set of mistakes awaiting the classically trained on their first data mining project! My advice? There is a lot to benefit from here. All of the "white box" walk through examples are great. Consider buying this book, the Pyle book Data Preparation for Data Mining (The Morgan Kaufmann Series in Data Management Systems), and Berry and Linoff Data Mining Techniques: For Marketing, Sales, and Customer Relationship Management, while skipping the other two in this trilogy. Use this book for the algorithm explanations, but be cautious otherwise. The screen shots and discussion of Clementine may be helpful to you, but note that Clementine 8.5 was used. DISCOVERING KNOWLEDGE IN DATA......Page 3 CONTENTS......Page 9 PREFACE......Page 13 1 INTRODUCTION TO DATA MINING......Page 19 What Is Data Mining?......Page 20 Need for Human Direction of Data Mining......Page 22 Cross-Industry Standard Process: CRISP–DM......Page 23 Case Study 1: Analyzing Automobile Warranty Claims: Example of the CRISP–DM Industry Standard Process in Action......Page 26 Fallacies of Data Mining......Page 28 Description......Page 29 Estimation......Page 30 Prediction......Page 31 Classification......Page 32 Clustering......Page 34 Association......Page 35 Case Study 2: Predicting Abnormal Stock Market Returns Using Neural Networks......Page 36 Case Study 3: Mining Association Rules from Legal Databases......Page 37 Case Study 4: Predicting Corporate Bankruptcies Using Decision Trees......Page 39 Case Study 5: Profiling the Tourism Market Using k-Means Clustering Analysis......Page 41 References......Page 42 Exercises......Page 43 Why Do We Need to Preprocess the Data?......Page 45 Data Cleaning......Page 46 Handling Missing Data......Page 48 Identifying Misclassifications......Page 51 Graphical Methods for Identifying Outliers......Page 52 Data Transformation......Page 53 Min–Max Normalization......Page 54 Z-Score Standardization......Page 55 Numerical Methods for Identifying Outliers......Page 56 Exercises......Page 57 Hypothesis Testing versus Exploratory Data Analysis......Page 59 Getting to Know the Data Set......Page 60 Dealing with Correlated Variables......Page 62 Exploring Categorical Variables......Page 63 Using EDA to Uncover Anomalous Fields......Page 68 Exploring Numerical Variables......Page 70 Exploring Multivariate Relationships......Page 77 Selecting Interesting Subsets of the Data for Further Investigation......Page 79 Binning......Page 80 Summary......Page 81 Exercises......Page 82 Data Mining Tasks in Discovering Knowledge in Data......Page 85 Statistical Approaches to Estimation and Prediction......Page 86 Univariate Methods: Measures of Center and Spread......Page 87 Statistical Inference......Page 89 Confidence Interval Estimation......Page 91 Bivariate Methods: Simple Linear Regression......Page 93 Dangers of Extrapolation......Page 97 Prediction Intervals for a Randomly Chosen Value of y Given x......Page 98 Multiple Regression......Page 101 Verifying Model Assumptions......Page 103 Exercises......Page 106 Supervised versus Unsupervised Methods......Page 108 Methodology for Supervised Modeling......Page 109 Bias–Variance Trade-Off......Page 111 Classification Task......Page 113 k-Nearest Neighbor Algorithm......Page 114 Distance Function......Page 117 Simple Unweighted Voting......Page 119 Weighted Voting......Page 120 Quantifying Attribute Relevance: Stretching the Axes......Page 121 k-Nearest Neighbor Algorithm for Estimation and Prediction......Page 122 Choosing k......Page 123 Exercises......Page 124 6 DECISION TREES......Page 125 Classification and Regression Trees......Page 127 C4.5 Algorithm......Page 134 Decision Rules......Page 139 Comparison of the C5.0 and CART Algorithms Applied to Real Data......Page 140 Exercises......Page 144 7 NEURAL NETWORKS......Page 146 Input and Output Encoding......Page 147 Simple Example of a Neural Network......Page 149 Sigmoid Activation Function......Page 152 Gradient Descent Method......Page 153 Back-Propagation Rules......Page 154 Example of Back-Propagation......Page 155 Learning Rate......Page 157 Momentum Term......Page 158 Sensitivity Analysis......Page 160 Application of Neural Network Modeling......Page 161 Exercises......Page 163 Clustering Task......Page 165 Hierarchical Clustering Methods......Page 167 Single-Linkage Clustering......Page 168 Complete-Linkage Clustering......Page 169 Example of k-Means Clustering at Work......Page 171 Application of k-Means Clustering Using SAS Enterprise Miner......Page 176 References......Page 179 Exercises......Page 180 Self-Organizing Maps......Page 181 Kohonen Networks......Page 183 Example of a Kohonen Network Study......Page 184 Application of Clustering Using Kohonen Networks......Page 188 Interpreting the Clusters......Page 189 Cluster Profiles......Page 193 Using Cluster Membership as Input to Downstream Data Mining Models......Page 195 Exercises......Page 196 Affinity Analysis and Market Basket Analysis......Page 198 Data Representation for Market Basket Analysis......Page 200 Support, Confidence, Frequent Itemsets, and the A Priori Property......Page 201 How Does the A Priori Algorithm Work (Part 1)? Generating Frequent Itemsets......Page 203 How Does the A Priori Algorithm Work (Part 2)? Generating Association Rules......Page 204 Extension from Flag Data to General Categorical Data......Page 207 J-Measure......Page 208 Application of Generalized Rule Induction......Page 209 When Not to Use Association Rules......Page 211 Do Association Rules Represent Supervised or Unsupervised Learning?......Page 214 Local Patterns versus Global Models......Page 215 Exercises......Page 216 11 MODEL EVALUATION TECHNIQUES......Page 218 Model Evaluation Techniques for the Estimation and Prediction Tasks......Page 219 Error Rate, False Positives, and False Negatives......Page 221 Misclassification Cost Adjustment to Reflect Real-World Concerns......Page 223 Decision Cost/Benefit Analysis......Page 225 Lift Charts and Gains Charts......Page 226 Interweaving Model Evaluation with Model Building......Page 229 Confluence of Results: Applying a Suite of Models......Page 230 Exercises......Page 231 EPILOGUE: “WE’VE ONLY JUST BEGUN”......Page 233 INDEX......Page 235 Annotation Learn Data Mining By Doing Data Miningdata Mining Can Be Revolutionary-but Only When It's Done Right. The Powerful Black Box Data Mining Software Now Available Can Produce Disastrously Misleading Results Unless Applied By A Skilled And Knowledgeable Analyst. Discovering Knowledge In Data: An Introduction To Data Mining Provides Both The Practical Experience And The Theoretical Insight Needed To Reveal Valuable Information Hidden In Large Data Sets. Employing A White Box Methodology And With Real-world Case Studies, This Step-by-step Guide Walks Readers Through The Various Algorithms And Statistical Structures That Underlie The Software And Presents Examples Of Their Operation On Actual Large Data Sets. Principal Topics Include:* Data Preprocessing And Classification* Exploratory Analysis* Decision Trees* Neural And Kohonen Networks* Hierarchical And K-means Clustering* Association Rules* Model Evaluation Techniquescomplete With Scores Of Screenshots And Diagrams To Encourage Graphical Learning, Discovering Knowledge In Data: An Introduction To Data Mining Gives Students In Business, Computer Science, And Statistics As Well As Professionals In The Field The Power To Turn Any Data Warehouse Into Actionable Knowledge. An Instructor's Manual Presenting Detailed Solutions To All The Problems In The Book Is Available Online. An Introduction To Data Mining -- Data Preprocessing -- Exploratory Data Analysis -- Statistical Approaches To Estimation And Prediction -- K-nearest Neighbor Algorithm -- Decision Trees -- Neural Networks -- Hierarchical And K-means Clustering -- Kohonen Networks -- Association Rules -- Model Evaluation Techniques. Daniel T. Larose. Includes Bibliographical References And Index. Also Available In An Electronic Version. Mode Of Access: World Wide Web. Learn Data Mining by doing data mining Data mining can be revolutionary-but only when it's done right. The powerful black box data mining software now available can produce disastrously misleading results unless applied by a skilled and knowledgeable analyst. Discovering Knowledge in Data: An Introduction to Data Mining provides both the practical experience and the theoretical insight needed to reveal valuable information hidden in large data sets. Employing a "white box" methodology and with real-world case studies, this step-by-step guide walks readers through the various algorithms and statistical structures that underlie the software and presents examples of their operation on actual large data sets. Principal topics include: Data preprocessing and classification Exploratory analysis Decision trees Neural and Kohonen networks Hierarchical and k-means clustering Association rules Model evaluation techniques Complete with scores of screenshots and diagrams to encourage graphical learning, Discovering Knowledge in Data: An Introduction to Data Mining gives students in Business, Computer Science, and Statistics as well as professionals in the field the power to turn any data warehouse into actionable knowledge. An Instructor's Manual presenting detailed solutions to all the problems in the book is available online. "Data mining can be revolutionary- but only when it's done right. The powerful black box data mining software now available can produce disastrously misleading results unless applied by a skilled and knowledgeable analyst. Discovering Knowledge in Data: An Introduction to Data Mining provides both the practical experience and the theoretical insight needed to reveal valuable information hidden in large data sets. Employing a "white box" methodology and with real-world case studies, this step-by-step guide walks readers through the various algorithms and statistical structures that underlie the software and presents examples of their operation on actual large data sets. Principal topics include: Data preprocessing and classification; Exploratory analysis; Decision trees; Neural and Kohonen networks; Hierarchical and k-means clustering; Association rules; Model evaluation techniques. Complete with scores of screenshots and diagrams to encourage graphical learning, Discovering Knowledge in Data: An Introduction to Data Mining gives students in Business, Computer Science, and Statistics as well as professionals in the field the power to turn any data warehouse into actionable knowledge."--Page 4 of cover Teach Like a Champion 2.0 is a complete update to the international bestseller. This teaching guide is a must-have for new and experienced teachers alike. Over 700,000 teachers around the world already know how the techniques in this book turn educators into classroom champions. With ideas for everything from classroom management to inspiring student engagement, you will be able to perfect your teaching practice right away. "Complete with scores of screenshots and diagrams to encourage graphical learning, Discovering Knowledge in Data: An Introduction to Data Mining gives students in Business, Computer Science, and Statistics as well as professionals in the field the power to turn any data warehouse into actionable knowledge."--Jacket