Modelling for Prediction vs. Modelling for Understanding: Commentary on Musso et al. (2013)
19 Dec 2013-Vol. 1, Iss: 2, pp 99-101
TL;DR: It is concluded that ANNs have high potential for theoretical and practical improvements in learning sciences and researchers in the learning sciences should prefer more theory-driven and parsimonious modelling techniques over ANNs whenever possible.
Abstract: Musso et al. (2013) predict students’ academic achievement with high accuracy one year in advance from cognitive and demographic variables, using artificial neural networks (ANNs). They conclude that ANNs have high potential for theoretical and practical improvements in learning sciences. ANNs are powerful statistical modelling tools but they can mainly be used for exploratory modelling. Moreover, the output generated from ANNs cannot be fully translated into a meaningful set of rules because they store information about input-output relations in a complex, distributed, and implicit way. These problems hamper systematic theory-building as well as communication and justification of model predictions in practical contexts. Modern-day regression techniques, including (Bayesian) structural equation models, have advantages similar to those of ANNs but without the drawbacks. They are able to handle numerous variables, non-linear effects, multi-way interactions, and incomplete data. Thus, researchers in the learning sciences should prefer more theory-driven and parsimonious modelling techniques over ANNs whenever possible.
Citations
More filters
01 Jan 2014
TL;DR: This commentary to the recent article by Musso et al. (2013) discusses issues related to model fitting, comparison of classification accuracy of generative and discriminative models, and two (or more) cultures of data modeling.
Abstract: This commentary to the recent article by Musso et al. (2013) discusses issues related to model fitting, comparison of classification accuracy of generative and discriminative models, and two (or more) cultures of data modeling. We start by questioning the extremely high classification accuracy with an empirical data from a complex domain. There is a risk that we model perfect nonsense perfectly. Our second concern is related to the relevance of comparing multilayer perceptron neural networks and linear discriminant analysis classification accuracy indices. We find this problematic, as it is like comparing apples and oranges. It would have been easier to interpret the model and the variable (group) importance’s if the authors would have compared MLP to some discriminative classifier, such as group lasso logistic regression. Finally, we conclude our commentary with a discussion about the predictive properties of the adopted data modeling approach.
1 citations
01 Jan 2018
TL;DR: This chapter describes one of some relatively new research methods in business, which are non-typical, non-statistical in nature, which rely on discovering unobserved or unnoticed patterns in the already available data and data sources.
Abstract: After discussing statistical techniques for data selection, collection, coding, manipulation, summarizing and presentation, this chapter describes one of some relatively new research methods in business, which are non-typical, non-statistical in nature. Artificial Neural Networks (ANNs), case-based reasoning, fuzzy logic and genetic algorithms are advanced techniques that show promises as enablers to solve some difficulties that may lie in analyzing and synthesizing complex systems, which include large quantities of data from several different sources into a coherent research model. Raising the idea up of discovering un-noticed observations or data in front of a researcher is for a purpose. One of the new techniques proposed in this chapter, like data mining, rely on discovering unobserved or unnoticed patterns in the already available data and data sources. This chapter will focus on using ANN method, what is it, who will use it, why and how to use it. The chapter ends by presenting the future trend in using this method, which is the combination among typical and non-typical methods.
Journal Article•
TL;DR: The results of applying the statistical Chi-square to contingency tables suggest a group of aspects that more mainly influence in achieve successful in the subjects Logic of Programming and Imperative Programming.
Abstract: The low academic performance at the university is influenced by many factors manifested mainly in the early years of study. However, the influential factors and the results presented in previous studies are different depending on the kind of universities. This work presents an approach to identify a group of factors or variables affecting the low academic performance of students in Computer Science in the university Agostinho Neto, of Angola, during first year of studies, specifically in two essential subjects of the course: Logic of Programming and Imperative Programming. From the psychological, pedagogical and social point of view the reasons to select those variables is analyzed and is compared by mean of a correlacional analysis for estimating the possible association between each independent variable and the variable criteria. Information employed belongs to three courses (about 800 students). The results of applying the statistical Chi-square to contingency tables suggest a group of aspects that more mainly influence in achieve successful in the subjects Logic of Programming and Imperative Programming.
Cites background from "Modelling for Prediction vs. Modell..."
...A partir de estudios de esta naturaleza se pueden elaborar modelos que ayuden a predecir qué alumnos están en riesgo de no tener éxito en algunas materias o año, desde el mismo inicio de su carrera, como se sugiere en [11]....
[...]
References
More filters
Book•
01 Jan 1991TL;DR: In this article, the effects of predictor scaling on the coefficients of regression equations are investigated. But, they focus mainly on the effect of predictors scaling on coefficients of regressions.
Abstract: Introduction Interactions between Continuous Predictors in Multiple Regression The Effects of Predictor Scaling on Coefficients of Regression Equations Testing and Probing Three-Way Interactions Structuring Regression Equations to Reflect Higher Order Relationships Model and Effect Testing with Higher Order Terms Interactions between Categorical and Continuous Variables Reliability and Statistical Power Conclusion Some Contrasts Between ANOVA and MR in Practice
27,897 citations
TL;DR: In this article, multiple regression is used to test and interpret multiple regression interactions in the context of multiple-agent networks. But it is not suitable for single-agent systems, as discussed in this paper.
Abstract: (1994). Multiple Regression: Testing and Interpreting Interactions. Journal of the Operational Research Society: Vol. 45, No. 1, pp. 119-120.
13,068 citations
"Modelling for Prediction vs. Modell..." refers methods in this paper
...Like ANNs, modern regression techniques can account for non-linear relations (Bates & Watts, 2007) and complex interactions between variables (Aiken & West, 1991)....
[...]
Book•
01 Aug 1988
TL;DR: This book offers a balanced presentation of the theoretical, practical, and computational aspects of nonlinear regression and provides background material on linear regression, including the geometrical development for linear and nonlinear least squares.
Abstract: Wiley-Interscience Paperback Series The Wiley-Interscience Paperback Series consists of selected books that have been made more accessible to consumers in an effort to increase global appeal and general circulation. With these new unabridged softcover volumes, Wiley hopes to extend the lives of these works by making them available to future generations of statisticians, mathematicians, and scientists. "The authors have put together an extraordinary presentation of concepts and methods concerning the use and analysis of nonlinear regression models ...highly recommend[ed] ...for anyone needing to use and/or understand issues concerning the analysis of nonlinear regression models." -Technometrics "[This book] provides a good balance of relevant theory and application with many examples ...[and it] provides the most balanced approach to theory and application appropriate for a first course in nonlinear regression modeling for graduate statistics students." -Mathematical Reviews "[This book] joins a distinguished list of publications with a reputation for balancing technical rigor with readability, and theory with application. [It] upholds tradition ...[and is] a worthwhile reference for the marketing researcher with a serious interest in linear models. " -Journal of Marketing Research This book offers a balanced presentation of the theoretical, practical, and computational aspects of nonlinear regression and provides background material on linear regression, including the geometrical development for linear and nonlinear least squares. The authors employ real data sets throughout, and their extensive use of geometric constructs and continuing examples makes the progression of ideas appear very natural. The book also includes pseudocode for computing algorithms.
3,202 citations
"Modelling for Prediction vs. Modell..." refers methods in this paper
...Like ANNs, modern regression techniques can account for non-linear relations (Bates & Watts, 2007) and complex interactions between variables (Aiken & West, 1991)....
[...]
Book•
01 Jan 1989TL;DR: In this paper, the authors combine the theoretical foundations of intelligent problem-solving with data structures and algorithms needed for its implementation, including logic, rule, object and agent-based architectures, along with example programs written in LISP and PROLOG.
Abstract: From the Publisher:
Combines the theoretical foundations of intelligent problem-solving with he data structures and algorithms needed for its implementation. The book presents logic, rule, object and agent-based architectures, along with example programs written in LISP and PROLOG.
The practical applications of AI have been kept within the context of its broader goal: understanding the patterns of intelligence as it operates in this world of uncertainty, complexity and change.
The introductory and concluding chapters take a new look at the potentials and challenges facing artificial intelligence and cognitive science. An extended treatment of knowledge-based problem-solving is given including model-based and case-based reasoning.
Includes new material on:
Fundamentals of search, inference and knowledge representation
AI algorithms and data structures in LISP and PROLOG Production systems, blackboards, and meta-interpreters including planers, rule-based reasoners, and inheritance systems.
Machine-learning including ID3 with bagging and boosting, explanation based learning, PAC learning, and other forms of induction
Neural networks, including perceptrons, back propogation, Kohonen networks, Hopfield networks, Grossberg learning, and counterpropagation. Emergent and social methods of learning and adaptation, including genetic algorithms, genetic programming and artificial life.
Object and agent-based problem solving and other forms of advanced knowledge representation
1,166 citations
Book•
01 Jan 2012
TL;DR: This work focuses on the implementation of Structural Equation Modeling in R with the sem and OpenMx Packages and on the development of scale construction and development models for this and other applications.
Abstract: Part 1. Background. R. Hoyle, Introduction and Overview. R. Matsueda, Key Advances in the History of Structural Equation Modeling. M. Ho, S. Stark, O. Chernyshenko, Graphical Representation of Structural Equation Models Using Path Diagrams. K. Bollen, R. Hoyle, Latent Variables in Structural Equation Modeling. J. Pearl, The Causal Foundations of Structural Equation Modeling. D. Bandalos, P. Gagne, Simulation Methods in Structural Equation Modeling. Part 2. Fundamentals. R. Kline, Assumptions in Structural Equation Modeling. R. Hoyle, Model Specification in Structural Equation Modeling. D. Kenny, S. Milan, Identification: A Nontechnical Discussion of a Technical Issue. P. Lei, Q. Wu, Estimation in Structural Equation Modeling. T. Lee, L. Cai, R. MacCallum, Power Analysis for Tests of Structural Equation Models. M. Edwards, R. Wirth, C. Houts, N. Xi, Categorical Data in the Structural Equation Modeling Framework. S. West, A. Taylor, W. Wu, Model Fit and Model Selection in Structural Equation Modeling. C. Chou, J. Huh, Model Modification in Structural Equation Modeling. L. Williams, Equivalent Models: Concepts, Problems, Alternatives. Part 3. Implementation. P. Malone, J. Lubansky, Preparing Data for Structural Equation Modeling: Doing Your Homework. J. Graham, D. Coffman, Structural Equation Modeling with Missing Data. G. Hancock, M. Liu, Bootstrapping Standard Errors and Data-Model Fit Statistics in Structural Equation Modeling. B. Byrne, Choosing Structural Equation Modeling Computer Software: Snapshots of LISREL, EQS, Amos, and Mplus. J. Fox, J. Byrnes, S. Boker, M. Neale, Structural Equation Modeling in R with the sem and OpenMx Packages. A. Boomsma, R. Hoyle, A. Panter, The Structural Equation Modeling Research Report. Part 4. Basic Applications. T. Brown, M. Moore, Confirmatory Factor Analysis. R. Millsap, M. Olivera-Aguilar, Investigating Measurement Invariance Using Confirmatory Factor Analysis. S. Green, M. Thompson, A Flexible Structural Equation Modeling Approach for Analyzing Means. J. Cheong, D. MacKinnon, Mediation/Indirect Effects in Structural Equation Modeling. H. Marsh, Z. Wen, B. Nagengast, K. Hau, Structural Equation Models of Latent Interaction. J. Biesanz, Autoregressive Longitudinal Models. T. Raykov, Scale Construction and Development Using Structural Equation Modeling. Part 5. Advanced Applications. J. Bovaird, N. Koziol, Measurement Models for Ordered-Categorical Indicators. S. Rabe-Hesketh, A. Skrondal, X. Zheng, Multilevel Structural Equation Modeling. M. Shiyko, N. Ram, K. Grimm, An Overview of Growth Mixture Modeling: A Simple Nonlinear Application in OpenMx. J. McArdle, Latent Curve Modeling of Longitudinal Growth Data. P. Wood, Dynamic Factor Models for Longitudinally Intensive Data: Description and Estimation via Parallel Factor Models of Cholesky Decomposition. D. Cole, Latent Trait-State Models. E. Ferrer, H. Song, Longitudinal Structural Models for Assessing Dynamics in Dyadic Interactions. S. Franic, C. Dolan, D. Borsboom, D. Boomsma, Structural Equation Modeling in Genetics. A. McIntosh, A. Protzner, Structural Equation Models of Imaging Data. D.Kaplan, S. Depaoli, Bayesian Structural Equation Modeling. M. Wall, Spatial Structural Equation Modeling. G. Marcoulides, M. Ing, Automated Structural Equation Modeling Strategies.
855 citations