Search results
1 – 10 of over 50000R. Dale Wilson and Harriette Bettis-Outland
Artificial neural network (ANN) models, part of the discipline of machine learning and artificial intelligence, are becoming more popular in the marketing literature and in…
Abstract
Purpose
Artificial neural network (ANN) models, part of the discipline of machine learning and artificial intelligence, are becoming more popular in the marketing literature and in marketing practice. This paper aims to provide a series of tests between ANN models and competing predictive models.
Design/methodology/approach
A total of 46 pairs of models were evaluated in an objective model-building environment. Either logistic regression or multiple regression models were developed and then were compared to ANN models using the same set of input variables. Three sets of B2B data were used to test the models. Emphasis also was placed on evaluating small samples.
Findings
ANN models tend to generate model predictions that are more accurate or the same as logistic regression models. However, when ANN models are compared to multiple regression models, the results are mixed. For small sample sizes, the modeling results are the same as for larger samples.
Research limitations/implications
Like all marketing research, this application is limited by the methods and the data used to conduct the research. The findings strongly suggest that, because of their predictive accuracy, ANN models will have an important role in the future of B2B marketing research and model-building applications.
Practical implications
ANN models should be carefully considered for potential use in marketing research and model-building applications by B2B academics and practitioners alike.
Originality/value
The research contributes to the B2B marketing literature by providing a more rigorous test on ANN models using B2B data than has been conducted before.
Details
Keywords
Shiloh James Howland and Ross A. A. Larsen
Graduate students often come to statistics courses with varying levels of motivation and previous academic preparation. Within the statistics education literature, there is a…
Abstract
Graduate students often come to statistics courses with varying levels of motivation and previous academic preparation. Within the statistics education literature, there is a growing consensus to guide instructors who want to help their students gain the requisite statistical knowledge so they can conduct their own research and report their results accurately. Recommendations from the literature include using real data, showing worked-out example problems, and providing immediate feedback to allow students to reflect on the correct and incorrect decisions they made in their analyses. This chapter describes the use of expert decision models (EDMs) in two graduate-level statistics courses – multiple regression and structural equation modeling. Decision-Based learning is an effective way to support graduate students’ developing thinking about statistics. In both courses, the students encounter the EDM through a series of assignments which guides students through the process of specifying a statistical model, running that model in Statistical Package for the Social Sciences or Mplus, and interpreting the results. These assignments use real datasets whenever possible and are designed to expose students to various issues they may experience in their research (missing data, violations of assumptions, etc.) and to illustrate how an expert would have adapted to those issues to complete the analysis. The EDM, with its just-in-time, just-enough instruction, helps students navigate these obstacles through guided practice and allows them to develop the conditional knowledge to handle issues that will arise as they carry out their own research.
Details
Keywords
William McCluskey, Peadar Davis, Martin Haran, Michael McCord and David McIlhatton
The aim of this paper is to investigate the comparative performance of an artificial neural network (ANN) and several multiple regression techniques in terms of their predictive…
Abstract
Purpose
The aim of this paper is to investigate the comparative performance of an artificial neural network (ANN) and several multiple regression techniques in terms of their predictive accuracy and capability of being used within the mass appraisal industry.
Design/methodology/approach
The methodology first tested that the data set had neglected non‐linearity which suggested that a non‐linear modelling technique should be applied. Given the capability of ANNs to model non‐linear data, this technique was used along with an OLS regression model (baseline model) and two non‐linear multiple regression techniques. In addition, the models were evaluated in terms of predictive accuracy and their capability of use within the mass appraisal environment.
Findings
Previous studies which have compared the predictive performance of an ANN model against multiple regression techniques are inconclusive. Having superior predictive capability is important but equally important is whether the technique can be successfully employed for the mass appraisal of residential property. This research found that a non‐linear regression model had higher predictive accuracy than the ANN. Also the output of the ANN was not sufficiently transparent to provide an unambiguous appraisal model upon which predicted values could be defended against objections.
Research limitations/implications
The research provides an informative view as to the efficacy of ANN methodology within the real estate field. A number of issues have been raised on the applicability of ANN models within the mass appraisal environment.
Practical implications
This work demonstrates that ANNs whilst useful as a predictive tool have a limited practical role for the assessment of residential property values for property tax purposes.
Originality/value
The work has taken forward the debate on the usefulness of ANN techniques within the mass appraisal environment.
Details
Keywords
This paper aims to discuss the utilization of artificial neural networks (ANNs) and multiple regression method for estimating surface roughness in milling medium density…
Abstract
Purpose
This paper aims to discuss the utilization of artificial neural networks (ANNs) and multiple regression method for estimating surface roughness in milling medium density fiberboard (MDF) material with a parallel robot.
Design/methodology/approach
In ANN modeling, performance parameters such as root mean square error, mean error percentage, mean square error and correlation coefficients (R2) for the experimental data were determined based on conjugate gradient back propagation, Levenberg–Marquardt (LM), resilient back propagation, scaled conjugate gradient and quasi-Newton back propagation feed forward back propagation training algorithm with logistic transfer function.
Findings
In the ANN architecture established for the surface roughness (Ra), three neurons [cutting speed (V), feed rate (f) and depth of cut (a)] were contained in the input layer, five neurons were included in its hidden layer and one neuron was contained in the output layer (3-5-1).Trials showed that LM learning algorithm was the best learning algorithm for the surface roughness. The ANN model obtained with the LM learning algorithm yielded estimation training values R2 (97.5 per cent) and testing values R2 (99 per cent). The R2 for multiple regressions was obtained as 96.1 per cent.
Originality/value
The result of the surface roughness estimation model showed that the equation obtained from the multiple regressions with quadratic model had an acceptable estimation capacity. The ANN model showed a more dependable estimation when compared with the multiple regression models. Hereby, these models can be used to effectively control the milling process to reach a satisfactory surface quality.
Details
Keywords
Edward E. Rigdon, Christian M. Ringle and Marko Sarstedt
Alongside structural equation modeling (SEM), the complementary technique of partial least squares (PLS) path modeling helps researchers understand relations among sets of…
Abstract
Alongside structural equation modeling (SEM), the complementary technique of partial least squares (PLS) path modeling helps researchers understand relations among sets of observed variables. Like SEM, PLS began with an assumption of homogeneity – one population and one model – but has developed techniques for modeling data from heterogeneous populations, consistent with a marketing emphasis on segmentation. Heterogeneity can be expressed through interactions and nonlinear terms. Additionally, researchers can use multiple group analysis and latent class methods. This chapter reviews these techniques for modeling heterogeneous data in PLS, and illustrates key developments in finite mixture modeling in PLS using the SmartPLS 2.0 package.
Elena G. Popkova and Aleksei V. Bogoviz
The purpose of the work is to model disproportions in development of regional economy of Russia and to determine perspectives and recommendations for overcoming them and achieving…
Abstract
The purpose of the work is to model disproportions in development of regional economy of Russia and to determine perspectives and recommendations for overcoming them and achieving the balance of the economy. The applied methods are based on Popkova's methodology of calculation of “underdevelopment whirlpools,” which allows conducting dynamic modeling of disproportions in development of regional economy. The research is performed in three consecutive stages. At the first stage, the dynamic model of development of the Russia's regional economy is compiled with the help of the methodology of “underdevelopment whirlpools” in federal districts of the Russian Federation based on GDP per capita. At the second stage, the key factors of emergence of disproportions in development of the Russia's regional economy are determined and models of multiple regression of development of the Russia's regional economy are compiled. At the third stage, target parameters of the determined factors are set for reducing the “underdevelopment whirlpools” in the Russia's regional economy by automatized solution of the optimization task with application of the simplex method and recommendations for overcoming the disproportions in development of the Russia's regional economy are compiled. As a result, it is concluded that regional economy of Russia is not well-balanced, as it has deep structural disproportions. These disproportions are caused by insufficient attention to peculiarities of regional economic systems during development and implementation of regional strategies of state management of economy. For more precise accounting of the influence of the key factors of appearance of disproportions and highly-effective management of them for overcoming the “underdevelopment whirlpools,” the algorithm of overcoming the disproportions in development of the Russia's regional economy is developed by the authors, which envisages various managerial measures depending on peculiarities of each Russian region.
Details
Keywords
Fredrick Otieno Okuta, Titus Kivaa, Raphael Kieti and James Ouma Okaka
The housing market in Kenya continues to experience an excessive imbalance between supply and demand. This imbalance renders the housing market volatile, and stakeholders lose…
Abstract
Purpose
The housing market in Kenya continues to experience an excessive imbalance between supply and demand. This imbalance renders the housing market volatile, and stakeholders lose repeatedly. The purpose of the study was to forecast housing prices (HPs) in Kenya using simple and complex regression models to assess the best model for projecting the HPs in Kenya.
Design/methodology/approach
The study used time series data from 1975 to 2020 of the selected macroeconomic factors sourced from Kenya National Bureau of Statistics, Central Bank of Kenya and Hass Consult Limited. Linear regression, multiple regression, autoregressive integrated moving average (ARIMA) and autoregressive distributed lag (ARDL) models regression techniques were used to model HPs.
Findings
The study concludes that the performance of the housing market is very sensitive to changes in the economic indicators, and therefore, the key players in the housing market should consider the performance of the economy during the project feasibility studies and appraisals. From the results, it can be deduced that complex models outperform simple models in forecasting HPs in Kenya. The vector autoregressive (VAR) model performs the best in forecasting HPs considering its lowest root mean squared error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE) and bias proportion coefficient. ARIMA models perform dismally in forecasting HPs, and therefore, we conclude that HP is not a self-projecting variable.
Practical implications
A model for projecting HPs could be a game changer if applied during the project appraisal stage by the developers and project managers. The study thoroughly compared the various regression models to ascertain the best model for forecasting the prices and revealed that complex models perform better than simple models in forecasting HPs. The study recommends a VAR model in forecasting HPs considering its lowest RMSE, MAE, MAPE and bias proportion coefficient compared to other models. The model, if used in collaboration with the already existing hedonic models, will ensure that the investments in the housing markets are well-informed, and hence, a reduction in economic losses arising from poor market forecasting techniques. However, these study findings are only applicable to the commercial housing market i.e. houses for sale and rent.
Originality/value
While more research has been done on HP projections, this study was based on a comparison of simple and complex regression models of projecting HPs. A total of five models were compared in the study: the simple regression model, multiple regression model, ARIMA model, ARDL model and VAR model. The findings reveal that complex models outperform simple models in projecting HPs. Nonetheless, the study also used nine macroeconomic indicators in the model-building process. Granger causality test reveals that only household income (HHI), gross domestic product, interest rate, exchange rates (EXCR) and private capital inflows have a significant effect on the changes in HPs. Nonetheless, the study adds two little-known indicators in the projection of HPs, which are the EXCR and HHI.
Details
Keywords
Jeffrey J. Burks, David W. Randolph and Jim A. Seida
This study examines the use of linear regressions that include interaction terms, finding frequent interpretation errors in published accounting research. We provide insights on…
Abstract
This study examines the use of linear regressions that include interaction terms, finding frequent interpretation errors in published accounting research. We provide insights on how to estimate, interpret, and present interactive regression models, and explain seldom-used but easily-implemented methods to report conditional marginal effects. We also examine the use of interaction terms in tax and financial reporting trade-off studies, evaluating the conceptual fit between a regression model with interactions and alternative definitions of trade-off. Although we advocate the use of interactive models, noise levels common in accounting research greatly reduce the ability to detect interaction effects.
Details
Keywords
Aims to commend SEM (structural equation modeling) that excels beyond multiple regression, which is a popular statistical technique to test the relationships of independent and…
Abstract
Aims to commend SEM (structural equation modeling) that excels beyond multiple regression, which is a popular statistical technique to test the relationships of independent and dependent variables, in expanding the explanatory ability and statistical efficiency for parsimonious model testing with a single comprehensive method. SEM is employed to find the real “best fitting” model. This article also presents an incremental approach to SEM, which is a procedural design and sounds workable for testing simple models and presents an example to test a parsimonious model of MBA knowledge and skills transfer using SEM and multiple regression. The results indicate that only one significant relationship can be justified by multiple regression. SEM, on the other hand, has helped to develop new relationships based on the modification indexes, which are also theoretically accepted. Finally, three relationships are shown to be significant and the “best fitting” structural model has been established.
Details
Keywords
Yuxin He, Yang Zhao and Kwok Leung Tsui
Exploring the influencing factors on urban rail transit (URT) ridership is vital for travel demand estimation and urban resources planning. Among various existing ridership…
Abstract
Purpose
Exploring the influencing factors on urban rail transit (URT) ridership is vital for travel demand estimation and urban resources planning. Among various existing ridership modeling methods, direct demand model with ordinary least square (OLS) multiple regression as a representative has considerable advantages over the traditional four-step model. Nevertheless, OLS multiple regression neglects spatial instability and spatial heterogeneity from the magnitude of the coefficients across the urban area. This paper aims to focus on modeling and analyzing the factors influencing metro ridership at the station level.
Design/methodology/approach
This paper constructs two novel direct demand models based on geographically weighted regression (GWR) for modeling influencing factors on metro ridership from a local perspective. One is GWR with globally implemented LASSO for feature selection, and the other one is geographically weighted LASSO (GWL) model, which is GWR with locally implemented LASSO for feature selection.
Findings
The results of real-world case study of Shenzhen Metro show that the two local models presented perform better than the traditional global model (OLS) in terms of estimation error of ridership and goodness-of-fit. Additionally, the GWL model results in a better fit than GWR with global LASSO model, indicating that the locally implemented LASSO is more effective for the accurate estimation of Shenzhen metro ridership than global LASSO does. Moreover, the information provided by both two local models regarding the spatial varied elasticities demonstrates the strong spatial interpretability of models and potentials in transport planning.
Originality/value
The main contributions are threefold: the approach is based on spatial models considering spatial autocorrelation of variables, which outperform the traditional global regression model – OLS – in terms of model fitting and spatial explanatory power. GWR with global feature selection using LASSO and GWL is compared through a real-world case study on Shenzhen Metro, that is, the difference between global feature selection and local feature selection is discussed. Network structures as a type of factors are quantified with the measurements in the field of complex network.
Details