Martin Natter

Conditional market segmentation by neural networks: a Monte-Carlo study

An artificial neural network (ANN) algorithm is proposed that incorporates both market segmentation and discriminant (regression) analysis of the segments. The method simultaneously estimates the models relating consumer characteristics to market segments, i.e., subjects are assigned to (unique) segments so that subjects within a class show similar purchase behavior and share the same characteristics (psychographics/sociodemographics). Parameters of all models are estimated by the backpropagation algorithm. The performance of the ANN methodology is assessed in a Monte-Carlo study. In contrast to the usual stepwise approach adopted in segmentation studies, our study found that simultaneous segmentation and discrimination are preferable for finding an overall optimum in that this way clusters are formed not only to create homogeneous submarkets but also to show a good dicriminatory behavior.

Evaluation of Aggressive Competitive Pricing Strategies

The main contribution of this paper is a method that allows one to study the effects of different degrees of competition. We find that optimal prices and profits are more sensitive to cooperative than to aggressive behavior on the part of competitors. With more aggressive policies, the average pricing level decreases and the average difference between high and low prices increases. An empirical model of the detergent market illustrates the methodology.

Profit Impacts of Aggressive and Cooperative Pricing Strategies

We study profit impacts of aggressive/cooperative pricing strategies in a dynamic oligopolistic environment. A logistic model with asymmetric reference prices is used to forecast market shares. Pricing strategies — optimized by simulated annealing — axe evaluated by simulating the empirical price distribution of competitors. It is shown that there are regions on the aggressiveness/cooperation path that a rationally operating manager would prefer to others, namely where his position is strongest as compared to the position of all rivals.

Ein sortimentsübergreifendes Entscheidungsunterstützungssystem für dynamische Preis- und Werbeplanung im DIY-Handel

Wir beschreiben ein Entscheidungsunterstutzungssystem zur dynamischen Preisund Werbeplanung. Das auf Wochendaten basierende Nachfragemodell berucksichtigt Preise, Referenzpreise, Saisonalitat, Artikelverfugbarkeit, Flugblatter und Rabatte. Wir quantifizieren Verbundeffekte und integrieren die daraus abgeleitete Gewinnsteigerung in das Preisoptimierungsmodell. Das Modell wurde bei einem osterreichischen Baumarkt entwickelt und implementiert. Aufgrund der praktischen Anforderungen wurde eine Zielfunktion verwendet, die die Strategie des Handelsunternehmens berucksichtigt. Acht Preisrunden mit Tausenden unterschiedlichen Artikeln wurden zur Evaluierung herangezogen. Unter Anwendung unterschiedlicher Vergleichswerte kann ein positiver Einfluss auf den Gewinn gezeigt werden. Das derzeit implementierte System steigerte den Bruttogewinn um durchschnittlich 8,1%, den Umsatz um 2,1%.

Specification and Estimation of Nonlinear Models with Dynamic Reference Prices

We specify reference prices as nonlinear dynamic latent variables represented as hidden units of an artificial neural net. Model parameters are estimated by an extended version of backpropagation. For a scanner data base containing prices and sales of seven brands of a consumer non-durable these models lead to better fits as compared to models that conceive reference prices as expectations. All reference price models are tested against a basic model with various functional forms. Results emphasize the importance of reference prices for the explanation of market share response.

A paper waste prediction model

The aim of this paper is to develop a model predicting the collected amount of waste paper at the regional level of municipalities. Leaming about the factors that influence the amount of collected paper is a prerequisite for the evaluation and reorganization of collection systems. We hypothesize that the amount of collected paper depends on both, the waste potential and factors which influence the convenience such as the density of collection sites. For this study, we use a sample of 649 municipalities. The data show a high variance in terms of the collected waste paper per person and year between the municipalities. We develop a multivariate regression model providing valuable insights about the relationship between demographic parameters and the amount of collected waste paper. Furthermore, in this novel approach we found a significant positive impact of the convenience of the collection system.

A Multilayer Perceptron for Clustering

We compare the performance of a specifically designed feedforward artificial neural network with one layer of hidden units to the K-means clustering technique using marketing resarch data. This proposed multilayer perceptron results in a two cluster solution that is confirmed by appropriate tests. On the other hand, the K-means algorithm fails in discovering any somewhat stronger cluster structure.

Clustering-Based Market Segmentation Using Neural Network Models

Most segmentation studies proceed in two steps, determining segments in the first step and looking for discriminating characteristics (segment descriptors) in the second step. Data-analytic methods usually applied in these steps are cluster analysis and discriminant analysis techniques, respectively. Artificial neural networks represent alternatives to better known statistical techniques. Certain types of artificial neural networks are closely related to well-known statistical methods. Principal components analysis, for example, is a special case of certain artificial neural network models. Therefore use of neural network models in market segmentation seems to be justified considering their greater generality.