P.V. (Sundar) Balakrishnan

A Study of the Classification Capabilities of Neural Networks Using Unsupervised Learning: A Comparison with K-Means Clustering

Several neural networks have been proposed in the general literature for pattern recognition and clustering, but little empirical comparison with traditional methods has been done. The results reported here compare neural networks using Kohonen learning with a traditional clustering method (K-means) in an experimental design using simulated data with known cluster solutions. Two types of neural networks were examined, both of which used unsupervised learning to perform the clustering. One used Kohonen learning with a conscience and the other used Kohonen learning without a conscience mechanism. The performance of these nets was examined with respect to changes in the number of attributes, the number of clusters, and the amount of error in the data. Generally, theK-means procedure had fewer points misclassified while the classification accuracy of neural networks worsened as the number of clusters in the data increased from two to five.

Comparative performance of the FSCL neural net and K-means algorithm for market segmentation

Abstract Given the success of neural networks in a variety of applications in engineering, such as speech and image quantization, it is natural to consider its application to similar problems in other domains. A related problem that arises in business is market segmentation for which clustering techniques are used. In this paper, we explore the ability of a specific neural network, namely the Frequency-Sensitive Competitive Learning Algorithm (FSCL), to cluster data for developing strategic marketing decisions. To this end, we investigate the comparative performance of FSCL vis-a-vis the K-means clustering technique. A cluster analysis conducted on brand choice data for the coffee category revealed that the two methodologies resulted in widely differing cluster solutions. In an effort to address the dispute over the appropriate methodology, a comparative performance investigation was undertaken using simulated data with known cluster solutions in a fairly large experimental design to mimic varying data quality to reflect data collection and measurement error. Based on the results of these studies, it is observed that a combination of the two methodologies, wherein the results of the FSCL network are input as seeds to the K-means, seems to provide more managerially insightful segmentation schemes.

Development of hybrid genetic algorithms for product line designs

In this paper, we investigate the efficacy of artificial intelligence (AI) based meta-heuristic techniques namely genetic algorithms (GAs), for the product line design problem. This work extends previously developed methods for the single product design problem. We conduct a large scale simulation study to determine the effectiveness of such an AI based technique for providing good solutions and bench mark the performance of this against the current dominant approach of beam search (BS). We investigate the potential advantages of pursuing the avenue of developing hybrid models and then implement and study such hybrid models using two very distinct approaches: namely, seeding the initial GA population with the BS solution, and employing the BS solution as part of the GA operators process. We go on to examine the impact of two alternate string representation formats on the quality of the solutions obtained by the above proposed techniques. We also explicitly investigate a critical managerial factor of attribute importance in terms of its impact on the solutions obtained by the alternate modeling procedures. The alternate techniques are then evaluated, using statistical analysis of variance, on a fairly large number of data sets, as to the quality of the solutions obtained with respect to the state-of-the-art benchmark and in terms of their ability to provide multiple, unique product line options.

Toward a Theory of Agenda Setting in Negotiations

To date, negotiation research in two-party situations has largely focused on single issues or on multiple issues bargained simultaneously. In this paper, we develop from a behavioral perspective, a conceptual framework and an associated set of propositions concerning the influence and interaction of a number of factors on agenda setting. We examine the consequences of negotiating multiple issues sequentially as opposed to discussing them simultaneously. Specifically, we posit (a) conditions under which sequential versus simultaneous negotiations are advantageous, (b) conditions that promote and inhibit integrative agreements between parties involved in sequential negotiations, and (c) conditions that foster greater utility and timeliness to the negotiating parties. In addition, directions for future research and methodological guidelines for testing the propositions are discussed. Copyright 1993 by the University of Chicago.

Efficiency evaluation of retail outlet networks

The authors generate alternative location-covering scenarios for retail outlet networks by using a programming model which guarantees the spatial market ‘adequacy’ of individual facility sites. The notion of relative spatial efficiency is then used to evaluate these scenarios and to develop information constructs which support many of the managerial decisions necessary for the appropriate structuring of such networks.

A maximin procedure for the optimal insertion timing of ad executions

It has been recommended that dominant brands should not suffer from long periods with little or no advertising. For such brands in a product category which is not subject to seasonality, it is important not to let the minimum effectiveness level fall too low. Toward this end, we present an analytical model for determining the optimal insertion timing pattern of a long run ad campaign which consists of a number of varying executions. We develop a computational procedure to calculate the time between the insertions within a pulse in order to maximize the minimum effectiveness level at any point in time. The focus of the procedure is to provide conceptual insights and on benchmark solutions, which can reduce the computational burden of simulations, rather than on providing an exact solution. We also provide a theoretical justification for the two common economic axioms of nonsatiation and diminishing marginal returns, in the context of an advertising model with exponential forgetting.

The Tempered Aspirations solution for bargaining problems with a reference point

Gupta and Livne (1988) modified Nash’s (1950) original bargaining problem through the introduction of a reference point restricted to lie in the bargaining set. Additionally, they characterized a solution concept for this augmented bargaining problem. We propose and axiomatically characterize a new solution concept for bargaining problems with a reference point: the Tempered Aspirations solution. In Kalai and Smorodinsky (1975), aspirations are given by the so called ideal or utopia point. In our setting, however, the salience of the reference point mutes or tempers the negotiators’ aspirations. Thus, our solution is defined to be the maximal feasible point on the line segment joining the modified aspirations and disagreement vectors. The Tempered Aspirations solution can be understood as a “dual” version of the Gupta–Livne solution or, alternatively, as a version of Chun and Thomson’s (1992) Proportional solution in which the claims point is endogenous. We also conduct an extensive axiomatic analysis comparing the Gupta–Livne to our Tempered Aspirations solution.

McThresh: Modeling Maximum Coverage with Threshold Constraints

Working within the location covering framework, the authors develop a mathematical programming approach for siting facilities which maximizes the extent of spatial market coverage, while maintaining threshold-demand levels for individual facility sites. The model, known as McTHRESH, can be formulated so that the number of facilities is either determined endogenously or stipulated exogenously. The working of the model is demonstrated with numerical examples.

Dual Objective Segmentation to Improve Targetability: An Evolutionary Algorithm Approach**

Cluster-based segmentation usually involves two sets of variables: (i) the needs-based variables (referred to as the bases variables), which are used in developing the original segments to identify the value, and (ii) the classification or background variables, which are used to profile or target the customers. The managers’ goal is to utilize these two sets of variables in the most efficient manner. Pragmatic managerial interests recognize the underlying need to start shifting from methodologies that obtain highly precise value-based segments but may be of limited practical use as they provide less targetable segments. Consequently, the imperative is to shift toward newer segmentation approaches that provide greater focus on targetable segments while maintaining homogeneity. This requires dual objective segmentation, which is a combinatorially difficult problem. Hence, we propose and examine a new evolutionary methodology based on genetic algorithms to address this problem. We show, based on a large-scale Monte Carlo simulation and a case study, that the proposed approach consistently outperforms the existing methods for a wide variety of problem instances. We are able to obtain statistically significant and managerially important improvements in targetability with little diminution in the identifiability of value-based segments. Moreover, the proposed methodology provides a set of good solutions, unlike existing methodologies that provide a single solution. We also show how these good solutions can be used to plot an efficient Pareto frontier. Finally, we present useful insights that would help managers in implementing the proposed solution approach effectively.

PRODLINE: Architecture of an Artificial Intelligence Based Marketing Decision Support System for PRODuct LINE Designs

Product line design is one of the most important decisions for an organization in today’s hypercompetitive world. Product line designs are NP-hard, which implies that it requires an unacceptable amount of time to obtain the guaranteed optimal solution to a problem of reasonable scale. Machine learning techniques such as genetic algorithms can provide very “good” solutions to these problems. In this chapter, we describe the architecture and user interface of a multi-feature decision support system, PRODLINE, which allows the decision maker to address the decision problem of product line designs. A key feature of the system is its ability to provide users with solutions using different solution techniques as well as the ability to change easily the algorithm parameters to assess if improvements in the solution are possible. A final novel and major advantage of the PRODLINE system is that it permits the user to consider strategic competitive responses to the optimal product line design problem.