Hierarchical Bayesian conjoint models incorporating measurement uncertainty

The authors explore situations where consumers supplement their judgments with a measurement of uncertainty about their own preferences, either implicitly or explicitly, and develop two sets of hierarchical Bayesian conjoint models incorporating such measurements. The first set of models uses the relative location of a rating to determine the importance or weight given to the rating, in a regression setting. The second set uses interval judgment as a dependent variable in a regression setting. After specifying the models, the authors perform a theoretical comparison with a basic Bayesian regression model. They show that, under different conditions, the proposed models will yield more precise individual-level partworth estimates. Two simulated data examples and data from a conjoint study are used to illustrate the gains that could be obtained from modeling uncertainty. In the empirical application, the authors show that model fit improves when ratings for items that respondents do not like are given more weight compared to ratings for items that they do like. Electronic Supplementary Material  The online version of this article (doi:) contains supplementary material, which is available to authorized users.

On the Similarity of Classical and Bayesian Estimates of Individual Mean Partworths

An exciting development in modeling has been the ability to estimate reliable individual-level parameters for choice models. Individual partworths derived from these parameters have been very useful in segmentation, identifying extreme individuals, and in creating appropriate choice simulators. In marketing, hierarchical Bayes models have taken the lead in combining information about the aggregate distribution of tastes with the individual's choices to arrive at a conditional estimate of the individual's parameters. In economics, the same behavioral model has been derived from a classical rather than a Bayesian perspective. That is, instead of Gibbs sampling, the method of maximum simulated likelihood provides estimates of both the aggregate and the individual parameters. This paper explores the similarities and differences between classical and Bayesian methods and shows that they result in virtually equivalent conditional estimates of partworths for customers. Thus, the choice between Bayesian and classical estimation becomes one of implementation convenience and philosophical orientation, rather than pragmatic usefulness.     

Assessment of potential retail segmentation variables An approach based on a subjective MCI resource allocation model

The application of a multiplicative competitive interaction (MCI) resource allocation model to assess potential segmentation variables in terms of their capacity to homogenise consumers’ patronage preferences is proposed. The method consists of grouping the potential customers by the variable in question, determining the shopping profile of each resulting segment, and comparing the results to identify insightful relationships between the variable and the shoppers’ retail selection criteria. An empirical test of the procedure in the context of the grocery retail market is subsequently presented and confirms the importance of evaluating easy-measurable demographic and socioeconomic variables as orientative indicators of shopping behaviour.     

Productive capabilities: An empirical analysis of their drivers

Recent contributions to the growth and trade literature have argued that the structure of an economy, as measured by its productive capabilities, is a key determinant for inter-country differences in development. Productive capabilities have been shown to be highly predictive of future economic growth, yet the country-level variables associated with them remain relatively unknown. In this paper, we empirically explore what variables are systematically associated with productive capabilities using a model averaging framework that can handle a very large number of potential explanatory variables without the need for arbitrary model selection. In order to estimate our dynamic panel specification, we propose a novel Bayesian averaging of classical estimates procedure based on the simple and efficient bias-corrected least squares dummy variable estimator. Our baseline and robustness analysis consider a large number of variables, sample periods and model priors. We find that there is persistence (as measured by the lagged dependent variable) and that variables, such as commodity terms of trade, energy availability, government consumption, capital per worker, arable land and capital inflows show a strong and robust association with capabilities.     

Bayesian Analysis of Simultaneous Demand and Supply

In models of demand and supply, consumer price sensitivity affects both the sales of a good through price, and the price that is set by producers and retailers. The relationship between the dependent variables (e.g., demand and price) and the common parameters (e.g., price sensitivity) is typically non-linear, especially when heterogeneity is present. In this paper, we develop a Bayesian method to address the computational challenge of estimating simultaneous demand and supply models that can be applied to both the analysis of household panel data and aggregated demand data. The method is developed within the context of a heterogeneous discrete choice model coupled with pricing equations derived from either specific competitive structures, or linear equations of the kind used in instrumental variable estimation, and applied to a scanner panel dataset of light beer purchases. Our analysis indicates that incorporating heterogeneity into the demand model all but eliminates the bias in the price parameter due to the endogeneity of price. The analysis also supports the use of a full information analysis.     

Response style corrected market segmentation for ordinal data

Survey data collected for market segmentation studies is typically ordinal in nature. As such, it is susceptible to response styles. Ignoring response styles can lead to market segments which do not differ in beliefs, but merely in how segment members use survey answer options and which possibly occur in addition to the belief segments. We propose a finite mixture model which simultaneously segments and corrects for response styles, permits heterogeneity in both beliefs and response styles, accommodates a range of different response styles, does not impose a certain relationship between the response style and belief segments, and is suitable for ordinal data. The performance of the model is tested using both artificial and empirical survey data.     

An efficient approach to quantile capital allocation and sensitivity analysis

In various fields of applications such as capital allocation, sensitivity analysis, and systemic risk evaluation, one often needs to compute or estimate the expectation of a random variable, given that another random variable is equal to its quantile at some prespecified probability level. A primary example of such an application is the Euler capital allocation formula for the quantile (often called the value‐at‐risk), which is of crucial importance in financial risk management. It is well known that classic nonparametric estimation for the above quantile allocation problem has a slower rate of convergence than the standard rate. In this paper, we propose an alternative approach to the quantile allocation problem via adjusting the probability level in connection with an expected shortfall. The asymptotic distribution of the proposed nonparametric estimator of the new capital allocation is derived for dependent data under the setup of a mixing sequence. In order to assess the performance of the proposed nonparametric estimator, AR‐GARCH models are proposed to fit each risk variable, and further, a bootstrap method based on residuals is employed to quantify the estimation uncertainty. A simulation study is conducted to examine the finite sample performance of the proposed inference. Finally, the proposed methodology of quantile capital allocation is illustrated for a financial data set.     

Identifying purchase-history sensitive shopper segments using scanner panel data and sequence alignment methods

Prior segmentation research has primarily focused on market behavior, lifestyle and socio-demographics rather than on purchase histories. In this article, the authors propose the use of sequence alignment methods to segment customers based on their purchase histories. The principles underlying the method are discussed, and the method is illustrated using scanner panel data. The findings suggest that, compared to the conventional methods, the proposed method results in a better segment solution that captures both the shopping-frequency and variety-seeking information. These findings have important implications for differentiation and market positioning strategies.     

Expectancy-value attitude models an analysis of critical measurement issues

Criteria for scaling beliefs and evaluations in the Fishbein model are considered, and a procedure is developed and illustrated for the proper test of multiplicative models. Hierarchical regression is shown to be a valid method for testing interaction hypotheses even when measures are only interval or ordinal scaled.     

Note: A new approach to the modeling of spatially dependent and heterogeneous geographical regions

We propose a new spatial modeling approach to calibrate the potential impact of spatial dependency and heterogeneity on the underlying drivers of customer service and/or satisfaction measurement. The newly proposed procedure derives regionally varying coefficients, provides more flexible fitting, improves calibration fit and predictive validation, and can potentially result in augmented managerial implications compared to existing procedures by utilizing a hierarchical Bayes framework with geographical boundary effects. Using synthetic datasets, we illustrate how the proposed model outperforms four relevant benchmark models including ordinary linear regression, a Spatially Dependent Segmentation model (Govind, Rabikar, and Mittal 2018), classic Geographically Weighted Regression, and Bayesian Geographically Weighted Regression. The improved performance is most prominent when there exist significant differences between geographic boundaries and/or irregular patterns of observation locations. In our automobile customer satisfaction application study, the proposed approach also demonstrates favorable performance compared to these benchmark models. We find a dramatically heterogeneous pattern regarding two covariates in the Mountain U.S. geographic division: dealership service is more important in urban areas (e.g., Phoenix, Salt Lake City and Denver) than in rural areas, but vice-versa concerning vehicle quality.     

Combining self-explicated priors with conjoint data using Bayesian regression

A Bayesian regression procedure (RBAYES) is proposed for the optimal combination of self-explicated data (priors) and conjoint judgments. The procedure does not require the design matrix for the conjoint judgments to be of full rank. The Bayesian regression procedure is similar to weighted least square in that it uses an information ratio to weight the priors. We provide empirical comparisons for the proposed method against (1) a Stein-type estimator (SBAYES) using one data set and (2) OLS applied to the data from an adaptive conjoint analysis using a second data set. In the second application we also use an alternating least squares procedure by itself and in combination with Bayesian regression (RBAYES+) to accommodate scale incompatibility as well as heteroscedasticity. In both applications we obtain superior results for the Bayesian regression procedure.     

A Cross-Validity Comparison of Conjoint Analysis and Choice Models at Different Levels of Aggregation

Several (ratings-based) conjoint analysis and experimental choice (choice-based conjoint) models are compared on their ability to predict both aggregate choice shares among the sample and individual choices in an availability validation task. While there was a weak relationship between validations at the individual and aggregate levels, several models stand out. In general, models capturing individual differences validated well at both the individual and aggregate level. The hierarchical Bayes choice and conjoint models validated particularly well.Among choice models, the hierarchical Bayes choice model had the highest aggregate and individual level-validations. It was followed by the hybrid and seven segment latent segment choice models. Overall, the highest validating ratings-based conjoint model was the hierarchical Bayes model. However, the seven segment latent segment conjoint model produced better aggregate choice share validations than any other conjoint model. These results indicate that validations can be improved either by using benefit segment models and/or merging different types of data to estimate more individualized models.In most cases, rescaling improved the ratings-based, but not the choice-based choice share validations. This suggests that one might adjust for differences between ratings and choice tasks before making choice share predictions.     

Conversion of ordinal attitudinal scales: An inferential Bayesian approach

The need for scale conversion may arise whenever an attitude of individuals is measured by independent entrepreneurs each using an ordinal scale of its own with possibly different numbers of (arbitrary) ordinal categories. Such situations are quite common in the marketing realm. The conversion of a score of an individual measured on one scale into an estimated score of a similar scale with a different range is the concern of this paper. An inferential Bayesian approach is adopted to analyze the situation where we believe the scale with fewer categories can be obtained by collapsing the finer scale. This leads to inferences concerning rules for the conversion of scales. Further, we propose a method for testing the validity of such a model. The use of the proposed methodology is exemplified on real data from surveys concerning performance evaluation and satisfaction.     

Estimating the extensive margin of trade

Understanding and quantifying the determinants of the number of sectors or firms exporting in a given country is of relevance for the assessment of trade policies. Estimation of models for the number of exporting sectors, however, poses a challenge because the dependent variable has both a lower and an upper bound, implying that the partial effects of the explanatory variables on the conditional mean of the dependent variable cannot be constant. We argue that ignoring these bounds can lead to erroneous conclusions and propose a flexible specification that accounts for the doubly-bounded nature of the dependent variable. We empirically investigate the problem and the proposed solution, finding significant differences between estimates obtained with the proposed estimator and those obtained with standard approaches.     

融合改进FT显著性与Grabcut的图像目标分割算法

针对目前复杂度较大的图像中目标分割速度较慢、显著性边界分割不明确等问题,提出了一种融合改进的FT（Frequency-tuned）显著性检测与Grabcut的图像分割算法。该算法首先通过改进基于频率调谐的FT显著性检测方法得到图像中显著性较高的区域,并利用SLIC（Simple Linear Iterative Clustering）算法对显著图进行预处理得到超像素图,能够有效改善边界的分割效果,然后通过以图论GraphCut算法为基础改进的Grabcut算法建立高斯混合模型。为了提高算法效率,通过聚类以超像素代替原像素,并反复迭代高斯混合模型（Gaussian Mixed Model,GMM）参数,最后利用最大流最小割算法得到最优目标分割结果。实验结果表明所提算法能够更准确更高效率地分割图像中的显著性目标,对高分辨率图像也有很好的适用效果,相比于其他算法在分割精度上提高10%左右,并具有较高的分割效率。     

Customer Lifetime Value as the Basis of Customer Segmentation

SUMMARY

In this article, we examine current trends in customer life-time value and customer segmentation models and identify key issues for future research. CLV-based segmentation is a segmentation approach that groups customers into meaningful segments based upon customer lifetime value and (potentially) other factors. In the article, we discuss the extent to which CLV-based segmentation meets the criteria for effective segmentation. We also identify six areas for future research: (1) models and management of “micro-segments,” (2) using CLV-based segmentation to improve the efficiency of marketing programs, (3) the need for more dynamic CLV-based segmentation models, (4) applying CLV-based customer segmentation to new products and new customers, (5) challenges associated with implementing CLV-based segmentation, and (6) the need for new models that enable firms to segment customers by response to marketing activities and CLV at different points in the customer decision process.     

Market segmentation with choice-based conjoint analysis

Choice-based conjoint analysis has increased in popularity in recent years among marketing practitioners. The typical practice is to estimate choice-based conjoint models at the aggregate level, given insufficient data for individual-level estimation of part-worths. We discuss a method for market segmentation with choice-based conjoint models. This method determines the number of market segments, the size of each market segment, and the values of segment-level conjoint part-worths using commonly collected conjoint choice data. A major advantage of the proposed method is that current (incomplete) data collection approaches for choice-based conjoint analysis can still be used for market segmentation without having to collect additional data. We illustrate the proposed method using commercial conjoint choice data gathered in a new concept test for a major consumer packaged goods company. We also compare the proposed method with ana priori segmentation approach based on individual choice frequencies.     

Using Segmentation and Prototyping in Engaging Politically-Salient Climate-Change Household Segments

Targeting messages to the different segments of a population is necessary to achieve support for policy addressing climate change. Finer segmentation and archetypal prototyping may be advantageous to provide an in-depth understanding of the most politically-salient segments. The research, conducted in Australia, used quantitative analysis to identify subsegments and prototypical respondents, followed by Jungian-style in-depth interviews to reveal the responses of segment representatives to different marketing stimuli. The results suggest that there are challenges in achieving majority support for action against climate change, but there are archetypal words and images that may garner action.