A maximum entropy approach to estimation and inference in dynamic models or Counting fish in the sea using maximum entropy

https://doi.org/10.1016/0165-1889(95)00864-0Get rights and content

Abstract

In this paper we consider estimation problems based on dynamic discrete time models. The first problem involves noisy state observations, where the state equation and the observation equation are nonlinear. The objective is to estimate the unknown parameters of the state and observation equations and the unknown values of the state variable. Next we consider the problem of estimating the parameters of the objective function and of the state equation in a linear-quadratic control problem. In each case, given time series observations, we suggest a nonlinear inversion procedure that permits the unknown underlying parameters to be estimated. Examples are presented to suggest the operational nature of the results.

References (24)

  • Fulton et al.

    Estimating the objectives of a public firm in a natural resource industry

    Journal of Environmental Economics and Management

    (1989)
  • L. Hansen et al.

    Formulating and estimating dynamic linear rational expectations models

    Journal of Economic Dynamics and Control

    (1980)
  • P. Berck et al.

    Estimating structural resource models when stock is uncertain: Theory and an application to pacific halibut

  • L.D. Broemeling

    Bayesian analysis of linear models

    (1985)
  • G. Chow

    Analysis and control of dynamic economic systems

    (1975)
  • G. Chow

    Econometric analysis by control methods

    (1981)
  • G. Chow

    Econometrics

    (1983)
  • I. Csiszár

    Why least squares and maximum entropy? An axiomatic approach to inference for linear inverse problems

    Annals of Statistics

    (1991)
  • A. Golan et al.

    Recovering information from multinomial response data

    Journal of the American Statistical Association

    (1996)
  • A. Jameson et al.

    Inverse problem of linear optimal control

    SIAM Journal of Control

    (1973)
  • E.T. Jaynes

    Information theory and statistical mechanics

    Physics Review

    (1957)
  • E.T. Jaynes

    Information theory and statistical mechanics II

    Physics Review

    (1957)
  • Cited by (47)

    • Driving mechanism of concentrated rural resettlement in upland areas of Sichuan Basin: A perspective of marketing hierarchy transformation

      2020, Land Use Policy
      Citation Excerpt :

      In the case of a small sample size, the estimation by classical methods (e.g., least squares) may provide parameter estimates with high variance. As a solution, the Generalized Maximum Entropy (GME) estimator was employed to deal with the problem of recovering information when the underling model is not completely known and the data are relatively limited, partial or incomplete (Golan et al., 1996). The GME estimator builds on the Shannon entropy-information measure, the classical maximum entropy principle and the generalized maximum entropy theory (Golan et al., 1997; Jaynes, 1957; Shannon, 1948).

    • Exploitable carrying capacity and potential biomass yield of sectors in the East China Sea, Yellow Sea, and East Sea/Sea of Japan large marine ecosystems

      2019, Deep-Sea Research Part II: Topical Studies in Oceanography
      Citation Excerpt :

      The HPM, which estimates ECC using surplus production models with time-series catch and fishing effort data, is simple and easy to apply, and requires much less data. A further advantage is that the non-equilibrium surplus production model's fit is reasonably good and the reliability of estimated parameters is relatively high compared with the EMM approach, since the parameters are usually estimated from a theoretically optimal range (Golan et al., 1996). This is the reason why many studies have used this approach (Garcia and Newton, 1997; Kim, 2016).

    • A preliminary study on ABC estimation approach for ecosystem-based TAC management

      2018, Ocean and Coastal Management
      Citation Excerpt :

      The parameters for estimation of biomass, yield, and fishing effort, such as intrinsic growth rate r = 0.37/year, carrying capacity K = 2,738,800, and catchability coefficient q = 2.58 × 10−5, were used. These parameters were assumed by the maximum entropy model (Golan et al., 1996), applying catch (mt) and fishing effort (haul) data of the Korean chub mackerel by large purse seine fishery from 1990 to 2008. Table 3 shows the input parameters for the prediction of reproductive potential and mean total length indicators.

    • The economics of spatial-dynamic processes: Applications to renewable resources

      2009, Journal of Environmental Economics and Management
    View all citing articles on Scopus
    View full text