Le modèle Multinomial Logit peut être interprété comme un cas particulier de perception multicouches à sortie Softmax, lorsque le perception est limité à une fonction d'utilité linéaire. Suffisamment flexible pour modéliser des choix résultant d'effets non-linéaires, sans spécifier a priori leur nature, l'approche neuronale ne souffre pas de biais de spécification. Les avantages respectifs des deux modèles étant complémentaires, ils peuvent être combinés pour tenir compte à la fois des non-linéarités potentielles et améliorer l'interprétabilité du modèle. Cette démarche est illustrée par un exemple artificiel et appliquée à des données lecture optique.
Get full access to this article
View all access options for this article.
References
1.
BaldiP.HornikK. (1989), Neural Networks and Principal Component Analysis: Learning from Examples without Local Minima, Neural Networks, 2, 1, 53–58.
2.
Ben AkivaM.LermanS. (1985), Discrete Choice Analysis, MIT Press.
3.
BentzY.MerunkaD. (1995), Modelling Brand Choice with the Multinomial Logit Model and Neural Networks: a Comparison of Methods and Results, in Actes des deuxièmes rencontres de la recherche neuronale en sciences économiques et de gestion, Université de Poitiers, 105–124.
4.
BishopC. M. (1995), Neural Networks for Pattern Recognition, Oxford University Press.
5.
BridleJ. (1990), Probabilistic Interpretation of Feedforward Classification Network Outputs, with Relationships to Statistical Pattern Recognition, Neuro-Computing: Algorithms, Architectures and Applications, New York: Springer.
6.
BurgessA.N. (1995), How Neural Networks Can Improve Database Marketing, The Journal of Database Marketing, 2, 312–327.
7.
BurgessA.N. (1995), Loan Risk Analysis using Neural Networks, The Journal of Targeting, Measurement and analysis for Marketing, 4, 24–37.
8.
DesmetP. (1993), Portées et limites de l'utilisation du modèle logit pour l'étude des comportements d'achat, Recherche et application en Marketing, 8, 3, 65–78.
9.
FaderP.S.HardieB. (1995), Modeling Consumer Choice Among SKU's, Marketing Working paper n° 95–603, London Business School.
10.
FriedmanJ.H. (1991), Multivariate Adaptative Regression Splines, Ann. Statist., 19, 1–141.
11.
GenschD.ReckerW. (1979), The Multinomial, Multiattribute Logit Choice Model, Journal of Marketing Research, 16, 1, 124–132.
12.
GuadagniP. M.LittleJ. D.C. (1983), A Logit Model of Brand Choice Calibrated on Scanner Data, Marketing Science, 2, 3, 203–238.
13.
HärdleW. (1990), Applied Nonparametric Regression, Cambridge, Cambridge University Press.
HastieT.TibshiraniR.BujaA. (1994), Flexible Discriminant Analysis by Optimal Scoring, Journal of the American Statistical Association, 89, 428, 1255–1270.
HauserJ.R. (1978), Testing the Accuracy, Usefulness, and Significance of Probabilistic Choice Models: an Information Theoretic Approach, Operations Research, 26, May, 406–421.
18.
HertzJ.KroghA.PalmerR. (1991), Introduction to the Theory of Neural Compution, Addison-Wesley.
19.
HopfieldJ. J. (1987), Learning Algorithms and Probability Distributions in Feed-forward and Feed-back Networks, in Proceedings of the National Academy of Sciences, 84, 8429–8433.
HruschkaH. (1993), Determining Market Response Functions by Neural Network Modeling: a Comparison to Econometric Techniques, European Journal of Operational Research, 66, 27–35.
22.
HruschkaH.NatterM. (1994), «Clustering-based Market Segmentation using Neural Networks», working Paper, revised version, March.
23.
KohonenT. (1988), Self-Organizing and Associative Memory, Second Edition: Springer-Verlag, New York.
24.
KumarA.RaoV.R.SoniH., (1995), An Empirical Comparison of Neural Networks and Logistic Regression Models, Marketing Letters, 6, 4, 251–263.
25.
LattinJ.BucklinE. (1989), Reference Effects of Price and Promotion on Brand Choice Behavior, Journal of Marketing Research, 26, 299–310.
26.
LattinJ. (1987), A Model of Balanced Choice Behavior, Marketing Science, 6, 1, 48–65.
27.
Le CunY.BoserB.DenkerJ.S.HendersonD.HowardR.E.HubbardW.JackelL.D. (1989), Backpropagation Applied to Handwritten Zip Code Recognition, Neural Computation, 1, 4, 541–551.
28.
MacFaddenD. (1974), Conditional Logit Analysis of Qualitative Choice Behavior, in ZarembkaP. (ed.), Frontiers of Econometrics, Academic Press.
29.
MalhotraN. (1984), «The Use of Linear Logit Models in Marketing Research», Journal of Marketing Research, 23, 1, 20–31.
30.
MeyerR.KahnB. (1991), Probabilistics Models of Consumer Choice Behavioral, Handbook of Consumer Behavior, Prentice-Hall Inc., 85–123.
31.
RefenesA.N. (1995), Neural Networks in the Capital Markets, Wiley Finance.
32.
RipleyB. (1994), Neural Networks and Related Methods for Classification, Journal of the Royal Statistical Society, 56, 3, 409–456.
33.
RumelhartD.E.McClellandJ.L. (1986), Parallel Distributed Processing, MIT Press.
34.
ShockerA.D.Ben-AkivaM.BoccaraB.NedungadiP. (1991), Consideration Set Influences on Consumer Decision- Making and Choice: Issues, Models and Suggestions, Marketing letters, 2, Août, 181–198.
35.
TellisG.J. (1988), Advertising Exposure, Loyalty and Brand Purchase: a Two-Stage Model of Choice, Journal of Marketing Research, 25, 134–144.
36.
TrippiR.TurbanE (1992), Neural Networks in Finance and Investing, Probus Publishing.
37.
VanhonackerW. (1993), What does the Multinomial Logit Model Really Measure?, Working Paper INSEAD 93/26/EPS/MKT/SM.
