Many supervised machine learning algorithms require a discrete feature space. In this paper, we review previous work on continuous feature discretization, identify defining characteristics of the methods, and conduct an empirical evaluation of several methods. We compare binning, an unsupervised discretization method, to entropy-based and purity-based methods, which are supervised algorithms. We found that the performance of the Naive-Bayes algorithm significantly improved when features were discretized using an entropy-based method. In fact, over the 16 tested datasets, the discretized version of Naive-Bayes slightly outperformed C4.5 on average. We also show that in some cases, the performance of the C4.5 induction algorithm significantly improved if features were discretized in advance; in our experiments, the performance never significantly degraded, an interesting phenomenon considering the fact that C4.5 is capable of locally discretizing features.

Classification is an important data mining problem. Although classification is a well-studied problem, most of the current classi-fication algorithms require that all or a por-tion of the the entire dataset remain perma-nently in memory. This limits their suitability for mining over large databases. We present a new decision-tree-based classification algo-rithm, called SPRINT that removes all of the memory restrictions, and is fast and scalable. The algorithm has also been designed to be easily parallelized, allowing many processors to work together to build a single consistent model. This parallelization, also presented here, exhibits excellent scalability as well. The combination of these characteristics makes the proposed algorithm an ideal tool for data min-ing. 1

Categories and Subject ���������� � �¨�������������������������¦���¦����������¡¤�� ¡ � ¡����������������¦¡¤����§�£����

Most statistical and machine-learning algorithms assume that the data is a random sample drawn from a station-ary distribution. Unfortunately, most of the large databases available for mining today violate this assumption. They were gathered over months or years, and the underlying pro-cesses generating them changed during this time, sometimes radically. Although a number of algorithms have been pro-posed for learning time-changing concepts, they generally do not scale well to very large databases. In this paper we propose an efficient algorithm for mining decision trees from continuously-changing data streams, based on the ultra-fast VFDT decision tree learner. This algorithm, called CVFDT, stays current while making the most of old data by growing an alternative subtree whenever an old one becomes ques-tionable, and replacing the old with the new when the new becomes more accurate. CVFDT learns a model which is similar in accuracy to the one that would be learned by reapplying VFDT to a moving window of examples every time a new example arrives, but with O(1) complexity per example, as opposed to O(w), where w is the size of the window. Experiments on a set of large time-changing data streams demonstrate the utility of this approach.

. Classification is an important problem in the emerging field of data mining. Although classification has been studied extensively in the past, most of the classification algorithms are designed only for memory-resident data, thus limiting their suitability for data mining large data sets. This paper discusses issues in building a scalable classifier and presents the design of SLIQ 1 , a new classifier. SLIQ is a decision tree classifier that can handle both numeric and categorical attributes. It uses a novel pre-sorting technique in the tree-growth phase. This sorting procedure is integrated with a breadth-first tree growing strategy to enable classification of disk-resident datasets. SLIQ also uses a new tree-pruning algorithm that is inexpensive, and results in compact and accurate trees. The combination of these techniques enables SLIQ to scale for large data sets and classify data sets irrespective of the number of classes, attributes, and examples (records), thus making it an ...

Algorithms for learning cIassification trees have had successes in ar-tificial intelligence and statistics over many years. This paper outlines how a tree learning algorithm can be derived using Bayesian statis-tics. This iutroduces Bayesian techniques for splitting, smoothing, and tree averaging. The splitting rule is similar to QuinIan’s information gain, while smoothing and averaging replace pruning. Comparative ex-periments with reimplementations of a minimum encoding approach, Quinlan’s C4 (1987) and Breiman et aL’s CART (1984) show the full Bayesian algorithm produces more accurate predictions than versions

In this doctoral dissertation, we study three basic problems in machine learning and two new hypothesis spaces with corresponding learning algorithms. The problems we investigate are: accuracy estimation, feature subset selection, and parameter tuning. The latter two problems are related and are studied under the wrapper approach. The hypothesis spaces we investigate are: decision tables with a default majority rule (DTMs) and oblivious read-once decision graphs (OODGs).

Many academic journals ask their authors to provide a list of about five to fifteen keywords, to appear on the first page of each article. Since these key words are often phrases of two or more words, we prefer to call them keyphrases. There is a wide variety of tasks for which keyphrases are useful, as we discuss in this paper. We approach the problem of automatically extracting keyphrases from text as a supervised learning task. We treat a document as a set of phrases, which the learning algorithm must learn to classify as positive or negative examples of keyphrases. Our first set of experiments applies the C4.5 decision tree induction algorithm to this learning task. We evaluate the performance of nine different configurations of C4.5. The second set of experiments applies the GenEx algorithm to the task. We developed the GenEx algorithm specifically for automatically extracting keyphrases from text. The experimental results support the claim that a custom-designed algorithm (GenEx)...

Classification of large datasets is an important data mining problem. Many classification algorithms have been proposed in the literature, but studies have shown that so far no algorithm uniformly outperforms all other algorithms in terms of quality. In this paper, we present a unifying framework for decision tree classifiers that separates the scalability aspects of algorithms for constructing a decision tree from the central features that determine the quality of the tree. This generic algorithm is easy to instantiate with specific algorithms from the literature (including C4.5, CART,

For large, real-world inductive learning problems, the number of training examples often must be limited due to the costs associated with procuring, preparing, and storing the data and/or the computational costs associated with learning from the data. One question of practical importance is: if n training examples are going to be selected, in what proportion should the classes be represented? In this article we analyze the relationship between the marginal class distribution of training data and the performance of classification trees induced from these data, when the size of the training set is fixed. We study twenty-six data sets and, for each, determine the best class distribution for learning. Our results show that, for a fixed number of training examples, it is often possible to obtain improved classifier performance by training with a class distribution other than the naturally occurring class distribution. For example, we show that to build a classifier robust to different misclassification costs, a balanced class distribution generally performs quite well. We also describe and evaluate a budgetsensitive progressive-sampling algorithm that selects training examples such that the resulting training set has a good (near-optimal) class distribution for learning.