## Dynamic Bayesian Networks for Classification of Business Cycles (1999)

Citations: | 2 - 0 self |

### BibTeX

@TECHREPORT{Sondhauss99dynamicbayesian,

author = {Ursula Sondhauss and Claus Weihs},

title = {Dynamic Bayesian Networks for Classification of Business Cycles},

institution = {},

year = {1999}

}

### OpenURL

### Abstract

We use Dynamic Bayesian networks to classify business cycle phases. We compare classifiers generated by learning the Dynamic Bayesian network structure on different sets of admissible network structures. Included are sets of network structures of the Tree Augmented Naive Bayes (TAN) classifiers of Friedman, Geiger, and Goldszmidt (1997) adapted for dynamic domains. The performance of the developed classifiers on the given data was modest.

### Citations

7440 |
Probabilistie Reasoning in Intelligent Systems: Networks of Plausible Inference
- Pearl
- 1988
(Show Context)
Citation Context ...GNP) alone and an analysis of the influence of these variables. Probabilistic inference in Bayesian networks can be used to answer diagnostic questions of influence and characterization of that kind (=-=Pearl, 1988-=-). It has also been shown that Bayesian networks can be used successfully for classification (Friedman, Geiger, and Goldszmidt, FGG, 1997). Dynamic Bayesian networks (Dean and Kanazawa 1988) can model... |

1618 |
Fundamentals of speech recognition
- L, Juang
- 1993
(Show Context)
Citation Context ...sition (S)TAN classifier. Particularly when \Pi C = C \Gamma1 these "transition TAN classifiers" have a structure that corresponds to the structure of the transition network of a Hidden Mark=-=ov model (Rabiner and Juang 1993-=-), when this is represented as a Dynamic Bayesian network with observation variable O := (A 1 ; :::; A n ) 0 . This might be suitable for many domains. In our domain this gives phases of business cycl... |

1132 | A Bayesian method for the induction of probabilistic networks from data - Cooper, Herskovits - 1992 |

948 | Learning bayesian networks: The combination of knowledge and statistical data - Heckerman, Geiger, et al. - 1995 |

533 |
Models of Business Cycles
- Lucas
- 1987
(Show Context)
Citation Context ... of other methods applied to the problem. 2.1 Variables The selection of predictor variables was carried out by our domain experts by means of both theoretical deliberations and the "stylized fac=-=ts" (Lucas 1983-=-) of the West German economy. A list of these variables is given in Table 1. In the following sections a vector of these variables or the set of them will be called stylized facts or Styfacts. For the... |

227 | Learning the structure of dynamic probabilistic networks - Friedman, Murphy, et al. - 1998 |

199 | Learning Bayesian belief networks. An approach based on the MDL principle
- Lam, Bacchus
- 1994
(Show Context)
Citation Context ...penalty in the scoring function for the complexity of B 2 B is needed. The scoring functions that are most frequently used to learn Bayesian networks fulfill this condition: the MDL scoring function (=-=Lam and Bacchus 1994-=-) and the BDe scoring function (Heckerman, Geiger, and Chickering 1995) combine the likelihood with some penalty relating to the complexity of the model. The MDL scoring metric is based on the princip... |

111 |
Probabilistic temporal reasoning
- Dean, Kanazawa
- 1988
(Show Context)
Citation Context ...ization of that kind (Pearl, 1988). It has also been shown that Bayesian networks can be used successfully for classification (Friedman, Geiger, and Goldszmidt, FGG, 1997). Dynamic Bayesian networks (=-=Dean and Kanazawa 1988-=-) can model the behaviour of business cycles as a multivariate time series. So in this paper we learn structures of Dynamic Bayesian networks on restricted sets of network structures (inspired by the ... |

96 | of Environmental Advisors - Council - 1996 |

66 |
Business-cycle phases and their transitional dynamics
- Filardo
- 1994
(Show Context)
Citation Context ...r the analysis of empirical developments and for testing complex systems of hypotheses. The new actuality of business cycle research can also be recognized by means of numerous publications (see e.g. =-=Filardo 1994-=-, Symposium on Developments in Business Cycle Research 1996) and also by the immediate discussion of new ideas in empirical business cycle analysis (see e.g. Council of Economic Advisers 1994). This w... |

47 | Multi-interval discretization of continuous valued attributes for classi learning - Fayyad, Irani - 1993 |

29 | Bayesian network classification with continuous attributes: getting the best of both discretization and parametric fitting - Friedman, Goldszmidt, et al. - 1998 |

19 | A multivariate discretization method for learning Bayesian networks from mixed data - Monti, Cooper - 1998 |

9 |
Konjunktur. Stilisierte Fakten, Theorie, Prognose. 2., völlig neu bearbeitete Ausgabe
- Tichy
- 1994
(Show Context)
Citation Context ...that. There are discussions among economists whether the quarters 1984/1 to 1986/1 are to be considered as the valley in the plateau of an upswing or whether two cycles should be classified (see e.g. =-=Tichy 1994-=-). In the quarters around the German reunification 1989 all our classifiers ran into problems. The following downswing from 1992/1 to 1994/1 is not captured by the STAN classifiers and the BN classifi... |

6 | West german business cycles 1963-1994: A multivariate discriminant analysis - Heilemann, Münch - 1996 |

2 | Lernen probabilistischer und possibilistischer Netze aus Daten: Theorie und Anwendung. Kuenstliche Intelligenz, Themenheft Data Mining 1, 11-17. Gesellschaft fuer Informatik - Borgelt, Kruse, et al. - 1998 |

1 |
Saisonbereinigung und makro oekonomische Modelle: Befunde und Erfahrungen mit dem RWI-Konjunkturmodell. ifo-Studien 41
- Heilemann, Barabas
- 1996
(Show Context)
Citation Context ...d 1955/4 to 1994/4 without missing values. The data is not seasonally adjusted, but this is balanced by the frequent use of growth rates with regard to the corresponding quarter of the previous year (=-=Heilemann and Barabas 1996-=-, p.404). These 157 observations were classified into business cycle phases by the domain experts based on economic and heuristic considerations. 3 Table 1: Predictor variables Abbrevation Variable GN... |

1 |
Das RWIKonjunkturmodell - Hypothesen, Struktur und Ergebnisse. RWI-Papiere 6
- Rau, Heilemann, et al.
- 1977
(Show Context)
Citation Context ... interests. 2 Domain and Data Researchers from the RWI developed an econometric model for predicting important aggregates of the national accounts and also for simulating important economic policies (=-=Rau et al. 1977-=-). Motivated by the increasing interest in the cyclical dimension of short term economic development, the RWI started looking at comparable cyclical phases in the past, when analyzing the current stag... |