Chp 4 Study Guide

What are the major application areas for data mining?

Applications are listed near the beginning of this section (pp. 160-161): CRM, banking, retailing and logistics, manufacturing and production, brokerage and securities trading, insurance, computer hardware and software, government and defense, travel, healthcare, medicine, entertainment, homeland security and law enforcement, and sports.

How do you think the discussion between privacy and data mining will progress? Why?

As technology advances and more information about people becomes easier to get, the privacy debate will adjust accordingly. People's expectations about privacy will become tempered by their desires for the benefits of data mining, from individualized customer service to higher security. As with all issues of social import, the privacy issue will include social discourse, legal and legislative decisions, and corporate decisions. The fact that companies often choose to self-regulate (e.g., by ensuring their data is de-identified) implies that we may as a society be able to find a happy medium between privacy and data mining.

Give examples of situations in which association would be an appropriate data mining technique.

Association rule mining is appropriate to use when the objective is to discover two or more items (or events or concepts) that go together.

List and briefly define the phases in the CRISP-DM process.

CRISP-DM provides a systematic and orderly way to conduct data mining projects. This process has six steps. First, an understanding of the data and an understanding of the business issues to be addressed are developed concurrently. Next, data are prepared for modeling; are modeled; model results are evaluated; and the models can be employed for regular use.

What is the major difference between cluster analysis and classification?

Classification methods learn from previous examples containing inputs and the resulting class labels, and once properly trained, they are able to classify future cases. Clustering partitions pattern records into natural segments or clusters.

Give examples of situations in which cluster analysis would be an appropriate data mining technique.

Cluster algorithms are used when the data records do not have predefined class identifiers (i.e., it is not known to what class a particular record belongs).

Define data mining. Why are there many different names and definitions for data mining?

Data mining is the process through which previously unknown patterns in data were discovered. Another definition would be "a process that uses statistical, mathematical, and artificial learning techniques to extract and identify useful information and subsequent knowledge from large sets of data." This includes most types of automated data analysis. A third definition: Data mining is the process of finding mathematical patterns from (usually) large sets of data; these can be rules, affinities, correlations, trends, or prediction models. Data mining has many definitions because it's been stretched beyond those limits by some software vendors to include most forms of data analysis in order to increase sales using the popularity of data mining.

What are the most common myths about data mining?

Data mining provides instant, crystal-ball predictions. Data mining is not yet viable for business applications. Data mining requires a separate, dedicated database. Only those with advanced degrees can do data mining. Data mining is only for large firms that have lots of customer data.

What are the most popular commercial data mining tools?

Examples of these vendors include IBM (IBM SPSS Modeler), SAS (Enterprise Miner), StatSoft (Statistica Data Miner), KXEN (Infinite Insight), Salford (CART, MARS, TreeNet, RandomForest), Angoss (KnowledgeSTUDIO, KnowledgeSeeker), and Megaputer (PolyAnalyst). Most of the more popular tools are developed by the largest statistical software companies (SPSS, SAS, and StatSoft).

What are the most popular free data mining tools? Why are they gaining overwhelming popularity (especially R)?

Probably the most popular free and open source data mining tool is Weka. Others include RapidMiner and Microsoft's SQL Server. Their popularity continues to grow because of their availability, features, and user communities. R remains very popular as a default language because of its feature base supporting data manipulation.

Give examples of situations in which association would be an appropriate data mining technique.

Sales transactions Credit card transactions Banking services Insurance service products Telecommunication services Medical records

What are the most common data mining mistakes/blunders? How can they be minimized and/or eliminated?

Selecting the wrong problem for data mining Ignoring what your sponsor thinks data mining is and what it really can and cannot do Leaving insufficient time for data preparation. It takes more effort than one often expects Looking only at aggregated results and not at individual records Being sloppy about keeping track of the mining procedure and results Ignoring suspicious findings and quickly moving on Running mining algorithms repeatedly and blindly. (It is important to think hard enough about the next stage of data analysis. Data mining is a very hands-on activity.) Believing everything you are told about data Believing everything you are told about your own data mining analysis Measuring your results differently from the way your sponsor measures them Ways to minimize these risks are basically the reverse of these items.

What are the major data mining processes?

Similar to other information systems initiatives, a data mining project must follow a systematic project management process to be successful. Several data mining processes have been proposed: CRISP-DM, SEMMA, and KDD.

What do you think are the reasons for these myths about data mining?

Students' answers will differ. Some answers might relate to fear of analytics, fear of the unknown, or fear of looking dumb.

Define Gini index. What does it measure?

The Gini index and information gain (entropy) are two popular ways to determine branching choices in a decision tree. The Gini index measures the purity of a sample. If everything in a sample belongs to one class, the Gini index value is zero.

What are some of the criteria for comparing and selecting the best classification technique?

The amount and availability of historical data The types of data, categorical, interval, ration, etc. What is being predicted—class or numeric value The purpose or objective

How does CRISP-DM differ from SEMMA?

The main difference between CRISP-DM and SEMMA is that CRISP-DM takes a more comprehensive approach—including understanding of the business and the relevant data—to data mining projects, whereas SEMMA implicitly assumes that the data mining project's goals and objectives along with the appropriate data sources have been identified and understood.

Give examples of situations in which classification would be an appropriate data mining technique. Give examples of situations in which regression would be an appropriate data mining technique.

Classification is for prediction that can be based on historical data and relationships, such as predicting the weather, product demand, or a student's success in a university. If what is being predicted is a class label (e.g., "sunny," "rainy," or "cloudy") the prediction problem is called a classification, whereas if it is a numeric value (e.g., temperature such as 68°F), the prediction problem is called a regression.

Why do you think the early phases (understanding of the business and understanding of the data) take the longest in data mining projects?

The early steps are the most unstructured phases because they involve learning. Those phases (learning/understanding) cannot be automated. Extra time and effort are needed upfront because any mistake in understanding the business or data will most likely result in a failed BI project.

What are some of the methods for cluster analysis?

The most commonly used clustering algorithms are k-means and self-organizing maps.

What are the main differences between commercial and free data mining software tools?

The main difference between commercial tools, such as Enterprise Miner and Statistica, and free tools, such as Weka and RapidMiner, is computational efficiency. The same data mining task involving a rather large dataset may take a whole lot longer to complete with the free software, and in some cases it may not even be feasible (i.e., crashing due to the inefficient use of computer memory).

Is data mining a new discipline? Explain.

Although the term data mining is relatively new, the ideas behind it are not. Many of the techniques used in data mining have their roots in traditional statistical analysis and artificial intelligence work done since the early part of the 1980s. New or increased use of data mining applications makes it seem like data mining is a new discipline. In general, data mining seeks to identify four major types of patterns: associations, predictions, clusters, and sequential relationships. These types of patterns have been manually extracted from data by humans for centuries, but the increasing volume of data in modern times has created a need for more automatic approaches. As datasets have grown in size and complexity, direct manual data analysis has increasingly been augmented with indirect, automatic data processing tools that use sophisticated methodologies, methods, and algorithms. The manifestation of such evolution of automated and semiautomated means of processing large datasets is now commonly referred to as data mining.

Identify at least three of the main data mining methods.

Classification learns patterns from past data (a set of information—traits, variables, features—on characteristics of the previously labeled items, objects, or events) in order to place new instances (with unknown labels) into their respective groups or classes. The objective of classification is to analyze the historical data stored in a database and automatically generate a model that can predict future behavior. Cluster analysis is an exploratory data analysis tool for solving classification problems. The objective is to sort cases (e.g., people, things, events) into groups, or clusters, so that the degree of association is strong among members of the same cluster and weak among members of different clusters. Association rule mining is a popular data mining method that is commonly used as an example to explain what data mining is and what it can do to a technologically less savvy audience. Association rule mining aims to find interesting relationships (affinities) between variables (items) in large databases.

Briefly describe the general algorithm used in decision trees. A general algorithm for building a decision tree is as follows:

Create a root node and assign all of the training data to it. Select the best splitting attribute. Add a branch to the root node for each value of the split. Split the data into mutually exclusive (non-overlapping) subsets along the lines of the specific split and mode to the branches. Repeat steps 2 and 3 for each and every leaf node until the stopping criteria is reached (e.g., the node is dominated by a single class label).

Why do you think the most popular tools are developed by statistics companies?

Data mining techniques involve the use of statistical analysis and modeling. So it's a natural extension of their business offerings.

What are the main data preprocessing steps? Briefly describe each step and provide relevant examples.

Data preprocessing is essential to any successful data mining study. Good data leads to good information; good information leads to good decisions. Data preprocessing includes four main steps (listed in Table 4.1 on page 167): data consolidation: access, collect, select and filter data data cleaning: handle missing data, reduce noise, fix errors data transformation: normalize the data, aggregate data, construct new attributes data reduction: reduce number of attributes and records; balance skewed data

What are the privacy issues in data mining?

Data that is collected, stored, and analyzed in data mining often contains information about real people. This includes identification, demographic, financial, personal, and behavioral information. Most of these data can be accessed through some third-party data providers. In order to maintain the privacy and protection of individuals' rights, data mining professionals have ethical (and often legal) obligations.

List and briefly define at least two classification techniques.

Decision tree analysis. Decision tree analysis (a machine-learning technique) is arguably the most popular classification technique in the data mining arena. Statistical analysis. Statistical classification techniques include logistic regression and discriminant analysis, both of which make the assumptions that the relationships between the input and output variables are linear in nature, the data is normally distributed, and the variables are not correlated and are independent of each other. Case-based reasoning. This approach uses historical cases to recognize commonalities in order to assign a new case into the most probable category. Bayesian classifiers. This approach uses probability theory to build classification models based on the past occurrences that are capable of placing a new instance into a most probable class (or category). Genetic algorithms. The use of the analogy of natural evolution to build directed search-based mechanisms to classify data samples. Rough sets. This method takes into account the partial membership of class labels to predefined categories in building models (collection of rules) for classification problems.

What recent factors have increased the popularity of data mining?

Following are some of the most pronounced reasons: More intense competition at the global scale driven by customers' ever-changing needs and wants in an increasingly saturated marketplace. General recognition of the untapped value hidden in large data sources. Consolidation and integration of database records, which enables a single view of customers, vendors, transactions, etc. Consolidation of databases and other data repositories into a single location in the form of a data warehouse. The exponential increase in data processing and storage technologies. Significant reduction in the cost of hardware and software for data storage and processing. Movement toward the de-massification (conversion of information resources into nonphysical form) of business practices.

What are some major data mining methods and algorithms?

Generally speaking, data mining tasks can be classified into three main categories: prediction, association, and clustering. Based on the way in which the patterns are extracted from the historical data, the learning algorithms of data mining methods can be classified as either supervised or unsupervised. With supervised learning algorithms, the training data includes both the descriptive attributes (i.e., independent variables or decision variables) as well as the class attribute (i.e., output variable or result variable). In contrast, with unsupervised learning the training data includes only the descriptive attributes. Figure 4.3 (p. 157) shows a simple taxonomy for data mining tasks, along with the learning methods, and popular algorithms for each of the data mining tasks.

What are the key differences between the major data mining methods?

Prediction: the act of telling about the future. It differs from simple guessing by taking into account the experiences, opinions, and other relevant information in conducting the task of foretelling. A term that is commonly associated with prediction is forecasting. Even though many believe that these two terms are synonymous, there is a subtle but critical difference between the two. Whereas prediction is largely experience and opinion based, forecasting is data and model based. That is, in order of increasing reliability, one might list the relevant terms as guessing, predicting, and forecasting, respectively. In data mining terminology, prediction and forecasting are used synonymously, and the term prediction is used as the common representation of the act. Classification: analyzing the historical behavior of groups of entities with similar characteristics, to predict the future behavior of a new entity from its similarity to those groups Clustering: finding groups of entities with similar characteristics Association: establishing relationships among items that occur together Sequence discovery: finding time-based associations Visualization: presenting results obtained through one or more of the other methods Regression: a statistical estimation technique based on fitting a curve defined by a mathematical equation of known type but unknown parameters to existing data Forecasting: estimating a future data value based on past data values