Gholamreza Nakhaeizadeh

Algorithms for association rule mining — a general survey and comparison

ABSTRACT Today there are several eAE ient algorithms that ope with the popular and omputationally expensive task of asso iation rule mining. A tually, these algorithms are more or less des ribed on their own. In this paper we explain the fundamentals of asso iation rule mining and moreover derive a general framework. Based on this we des ribe todays approa hes in ontext by pointing out ommon aspe ts and di eren es. After that we thoroughly investigate their strengths and weaknesses and arry out several runtime experiments. It turns out that the runtime behavior of the algorithms is mu h more similar as to be expe ted.

Cost-sensitive pruning of decision trees

The pruning of decision trees often relies on the classification accuracy of the decision tree. In this paper, we show how the misclassification costs, a related criterion applied if errors vary in their costs, can be integrated in several well-known pruning techniques.

Machine learning and statistics: the interface

Statistical Properties of Tree-Based Approaches to Classification The Decision Tree Algorithm CAL5 Based on a Statistical Approach to its Splitting Algorithm Probabilistic Symbolic Classifiers: An Empirical Comparison from a Statistical Perspective A Multistrategy Approach to Learning Multiple Dependent Concepts Quality of Decision Rules - Definition and Classification Schemes for Multiple Rules DIPOL - A Hybrid Piecewise Linear Classifier Combining Classification Procedures Distance-based Decision Trees Learning Fuzzy Controllers from Examples Some Developments in Statistical Credit Scoring Combination of Statistical and Other Learning Methods to Predict Financial Time Series.

Data Mining of Association Rules and the Process of Knowledge Discovery in Databases

In this paper we deal with association rule mining in the context of a complex, interactive and iterative knowledge discovery process. After a general introduction covering the basics of association rule mining and of the knowledge discovery process in databases we draw the attention to the problematic aspects concerning the integration of both. Actually, we come to the conclusion that with regard to human involvement and interactivity the current situation is far from being satisfying. In our paper we tackle this problem on three sides: First of all there is the algorithmic complexity. Although todays algorithms efficiently prune the immense search space the achieved run times do not allow true interactivity. Nevertheless we present a rule caching schema that significantly reduces the number of mining runs. This schema helps to gain interactivity even in the presence of extreme run times of the mining algorithms. Second, today the mining data is typically stored in a relational database management system. We present an efficient integration with modern database systems which is one of the key factors in practical mining applications. Third, interesting rules must be picked from the set of generated rules. This might be quite costly because the generated rule sets normally are quite large whereas the percentage of useful rules is typically only a very small fraction. We enhance the traditional association rule mining framework in order to cope with this situation.

Wissensentdeckung in Datenbanken und Data Mining: Ein überblick

Dieser Artikel gibt einen uberblick uber das Gebiet der Wissensentdeckung in Datenbanken und Data Mining. Ferner gibt der Artikel eine ubersicht zu existierenden Techniken, Werkzeugen und Anwendungen in wissenschaftlicher Forschung und industrieller Praxis. Die verschiedenen Phasen des Prozesses der Wissensentdeckung werden vorgestellt und analysiert. Es gibt eine Reihe von Data Mining Zielen, die sich durch Anwendung des extrahierten Wissens bearbeiteten lassen. Wir beschreiben diese Ziele und stellen die entsprechenden Verfahren vor, die zur Erreichung dieser Ziele geeignet sind. Solche Verfahren basieren auf statistischen Methoden, neuronalen Netzen, Case-Based Reasoning und symbolischen Lernverfahren. Einige wichtige Phasen des Prozesses, wie die Vorbereitung der Daten, die eigentliche Entdeckung neuen Wissens und Bewertung der Ergebnisse werden wir ausfuhrlicher diskutieren. Inzwischen hat die Wissensentdeckung in Datenbanken in verschiedenen Gebieten zahlreiche Anwendungen gefunden. Auserdem sind die Anzahl der existierenden Systeme fur die Wissensentdeckung explosionsartig in die Hohe gestiegen. Aus diesem Grund ist eine Beschreibung diverser Anwendungen und die Vorstellung aller existierender Systeme nicht moglich. Wir stellen jedoch einige Anwendungen vor und beschreiben einige ausgewahlte Systeme. Ein uberblick uber die aktuellen Forschungsthemen schliest den Artikel ab.

Incorporating Prior Knowledge About Financial Markets Through Neural Multitask Learning

We present the systematic method of Multitask Learning for incorporating prior knowledge (hints) into the inductive learning system of neural networks. Multitask Learning is an inductive transfer method which uses domain information about related tasks as inductive bias to guide the learning process towards better solutions of the main problem. These tasks are presented to the learning system in a shared representation. This paper argues that there exist many opportunities for Multitask Learning especially in the world of financial modeling: It has been shown, that many interdependencies exist between international financial markets, different market sectors and financial products. Models with an isolated view on a single market or a single product therefore ignore this important source of information. An empirical example of Multitask Learning is presented where learning additional tasks improves the forecasting accuracy of a neural network used to forecast the changes of five major German stocks.

The causality direction in consumption–income process and sensitivity to lag structures

Using the statistical procedures developed by Granger (1969) and Sims (1972), the causality direction in the consumption–income process is examined in this study. The results provide some justification in concluding that the causality direction is from income to consumption. It can also be seen that the results of causality tests are very sensitive to the lag lengths employed. The statistical procedure, which is based on Grangerss contribution, appears to be more consistent than Simss procedure.

Efficient Rule Retrieval and Postponed Restrict Operations for Association Rule Mining

Knowledge discovery in databases is a complex, iterative, and highly interactive process. When mining for association rules, typically interactivity is largely smothered by the execution times of the rule generation algorithms. Our approach is to accept a single, possibly expensive run, but all subsequent mining queries are supposed to be answered interactively by accessing a sophisticated rule cache. However there are two critical aspects. First, access to the cache must be efficient and comfortable. Therefore we enrich the basic association mining framework by descriptions of items through application dependent attributes. Furthermore we extend current mining query languages to deal with these attributes through ? and ? quantifiers. Second, the cache must be prepared to answer a broad variety of queries without rerunning the mining algorithm. A main contribution of this paper is that we show how to postpone restrict operations on the transactions from rule generation to rule retrieval from the cache. That is, without actually rerunning the algorithm, we efficiently construct those rules from the cache that would have been generated if the mining algorithm were run on only a subset of the transactions. In addition we describe how we implemented our ideas on a conventional relational database system. We evaluate our prototype concerning response times in a pilot application at DaimlerChrysler. It turns out to satisfy easily the demands of interactive data mining.

REVI-MINER, a KDD-environment for deviation detection and analysis of warranty and goodwill cost statements in automotive industry

REVI-MINER is a KDD-environment which supports the detection and analysis of deviations in warranty and goodwill cost statements. The system was developed within the framework of a cooperation between DaimlerChrysler Research & Technology and Global Service and Parts (GSP) and is based upon the CRISP-DM methodology as a widely accepted process model for the solution of Data Mining problems. Also, we have implemented different approaches based on Machine learning and statistics which can be utilized for data cleaning in the preprocessing phase. The Data Mining models applied have been developed by using a statistical deviation detection approach. The tool supports controllers in their task of auditing the authorized repair shops. In this paper we describe the development phases which have led to REVI-MINER.

WAPS, a data mining support environment for the planning of warranty and goodwill costs in the automobile industry

WAPS (WArranty Planning System) is a Data Mining support environment which is developed within a cooperation between DaimlerChrysler Research & Technology, Ulm and the Sales and Services Direction, Stuttgart. Data Mining modelling is performed in WAPS by using an innovative approach based on regression analysis taking into account the structural change of warranty and goodwill costs data. In this paper, we describe the different development phases of WAPS.