Young S. Kwon

Ordinal Pairwise Partitioning (OPP) Approach to Neural Networks Training in Bond rating

Statistical classification methods such as multivariate discriminant analysis have been widely used in bond rating classification in spite of the limitations of the methodology. Recently, neural networks have emerged as new methods for business classification. This approach to neural networks training is to categorize a new instance as one of the predefined bond classes. Such a conventional approach has limitations in dealing with the ordinal nature of bond rating. In addition, most of the prior studies have used sample data which are evenly divided among the classes. However, the natural population in real application is usually unevenly divided among the classes. Under such circumstances, it is hard to achieve good predictive performance. As the number of classes to be recognized increases, the predictive performance decreases. In this article, to increase the predictive performance in real-world bond rating, we propose the ordinal pairwise partitioning (OPP) approach to backpropagation neural networks training. The main idea of the OPP approach is to partition the data set in the ordinal and pairwise manner according to the output classes. Then, each backpropagation neural networks model is trained by using each partitioned data set and is separately used for classification. Experimental results show that the predictive performance of the proposed OPP approach can be significantly enhanced, when compared to the conventional neural networks modeling approach as well as multivariate discriminant analysis. The OPP approach has two computation methods, and we discuss under which circumstances one method performs better than the other. We also show the generalizability of the OPP approach.

A practical approach to bankruptcy prediction for small businesses: Substituting the unavailable financial data for credit card sales information

Small businesses are open to the elements of both consumer and business credit risks. One of the problems in bankruptcy prediction for small businesses is that the official financial data in most cases are not available for evaluating the business credit risks. In order to alleviate this problem, we propose to use the credit card sales information as a substitute for the official financial data in developing a bankruptcy prediction model. In most cases, the credit card sales information is available because most small businesses are member stores of credit card processors. First, we derived several variables using the credit card sales information, including business period, sales scale, sales fluctuation, sales pattern and business categorys bankruptcy ratio, etc. Then we developed support vector machines (SVM) based model. The empirical analyses show that credit card sales information is an acceptable substitute for the financial data in predicting the bankruptcy of small businesses. In addition, the proposed SVM model exhibits superior performance compared to other classifiers such as neural networks, CART, C5.0, multivariate discriminant analysis (MDA), and logistic regression analysis (LRA).

Performance Improvement of Bankruptcy Prediction using Credit Card Sales Information of Small & Micro Business

Due to the poor financial statements which represent credit risk of small and micro business, its been difficult to develop the credit evaluation model that reflects both consumer credit risk and business credit risk of small and micro business. The purpose of this study is to develop the credit evaluation model for small and micro businesses using credit card sales information in lieu of poor financial information. In order to develop the model, we derive some variables and analyze the relationship between good and bad credits. We find out that twelve variables are significant in predicting good or bad risk for small and micro business, which are categorized into the business period, scale for sale, a fluctuation in sales, sales pattern and business categorys bankruptcy ratio, etc. We employ the new statistical learning technique, support vector machines (SVM) as a classifier. We use grid search technique to find out better parameter for SVM. The experimental result shows that credit card sales information could be a good substitute for the financial data on business credit risk in predicting the bankruptcy for small-micro businesses. In addition, we also find out that SVM performs best, when compared with other classifiers such as neural networks, CART, C5.0, multivariate discriminant analysis (MDA), and logistic regression analysis (IRA).

Analyzing customer behavior from shopping path data using operation edit distance

Radio frequency identification (RFID) technology has been successfully applied to gather customers’ shopping habits from their motion paths and other behavioral data. The customers’ behavioral data can be used for marketing purposes, such as improving the store layout or optimizing targeted promotions to specific customers. Some data mining techniques, such as clustering algorithms can be used to discover customers’ hidden behaviors from their shopping paths. However, shopping path data has peculiar challenges, including variable length, sequential data, and the need for a special distance measure. Due to these challenges, traditional clustering algorithms cannot be applied to shopping path data. In this paper, we analyze customer behavior from their shopping path data by using a clustering algorithm. We propose a new distance measure for shopping path data, called the Operation edit distance, to solve the aforementioned problems. The proposed distance method enables the RFID customer shopping path data to be processed effectively using clustering algorithms. We have collected a real-world shopping path data from a retail store and applied our method to the dataset. The proposed method effectively determined customers’ shopping patterns from the data.

A Practical Approach to the Shopping Path Clustering

This paper proposes a new clustering approach for customer shopping paths. The approach is based on the Apriori algorithm and LCS Longest Common Subsequence algorithms. We devised new similarity and performance measurements for the clustering. In this approach, we do not require data normalization for preprocessing, which leads to an easy and practical application and implementation of the proposed approach. The experiment results show that the proposed approach performs well compared with k-medoids clustering.

Current Approaches in Applied Artificial Intelligence

Managing uncertainty in risk assessment is a crucial issue for better decision making and especially when it is adapted to the three standards ISO 9001, OHSAS 18001 and ISO 14001. This paper proposes a new risk assessment approach able to manage risk in the context of integrated management system (IMS-QSE) while taking into account the uncertainty characterizing the whole process. The proposed approach is mainly based on fuzzy set theory and Monte Carlo simulation to provide an appropriate risk estimation values and adequate decisions regarding the three management systems Quality, Security and Environment. In order to show the effectiveness of our approach, we have performed simulations on real database in the petroleum field at TOTAL TUNISIA

A New Approach to Spatial Pattern Clustering based on Longest Common Subsequence with application to a Grocery

Identifying the major moving patterns of shoppersʼ movements in the selling floor has been a longstanding issue in the retailing industry. With the advent of RFID technology, it has been easier to collect the moving data for a individual shopperʼs movement. Most of the previous studies used the traditional clustering technique to identify the major moving pattern of customers. However, in using clustering technique, due to the spatial constraint (aisle layout or other physical obstructions in the store), standard clustering methods are not feasible for moving data like shopping path should be adjusted for the analysis in advance, which is time-consuming and causes data distortion. To alleviate this problems, we propose a new approach to spatial pattern clustering based on longest common subsequence (LCSS). Experimental results using the real data obtained from a grocery in Seoul show that the proposed method performs well in finding the hot spot and dead spot as well as in finding the major path patterns of customer movements. 1) Keyword: customer path, shopping behavior, exploratory analysis, LCSS, RFID

Inductive learning performance changing with relevant inputs

One of the difficulties in using the current information retrieval systems is that it is hard for a user, especially a novice, to formulate a query effectively. One solution to this problem is to automate the process of query reformulation using the relevance feedback from the previous search. In this research, a Boolean query is viewed as a classifier and a decision tree classifier (ID3) is revised to act as a query in information retrieval (call it ID3-IR). The current emphasis in our experiments is to analyze the changes in the retrieval performance (measured by recall, precision, and E) of the ID3-IR using a different number of relevant input documents. Based on the test set, MEDLARS, it is shown that an input set with more relevant documents achieves higher recall and lower precision. In overall performance analysis measured by E, an input set with more relevant documents is superior to one with less relevant documents after the second reformulation.