Hierarchical generalized linear model

Overview and model

Model

In a hierarchical model, observations are grouped into clusters, and the distribution of an observation is determined not only by common structure among all clusters but also by the specific structure of the cluster where this observation belongs. So a random effect component, different for different clusters, is introduced into the model. Let y be the response, u be the random effect, g be the link function, \eta = X \beta , and v = v(u) is some strictly monotone function of u. In a hierarchical generalized linear model, the assumption on y|u and u need to be made: y\mid u \sim\ f(\theta,,\phi) and u \sim\ f_u(\alpha).

The linear predictor is in the form:

: g(E(y)) = g(\mu) = \eta = X \beta + v ,

where g is the link function, \mu = E(y) , \eta = X \beta + v , and v = v(u) is a monotone function of u. In this hierarchical generalized linear model, the fixed effect is described by \beta , which is the same for all observations. The random component u is unobserved and varies among clusters randomly. So v takes the same value for observations in the same cluster and different values for observations in different clusters.

Identifiability

Identifiability is a concept in statistics. In order to perform parameter inference, it is necessary to make sure that the identifiability property holds. In the model stated above, the location of v is not identifiable, since

: X \beta + v = (X \beta + a) + (v - a) ,

for constant a. In order to make the model identifiable, we need to impose constraints on parameters. The constraint is usually imposed on random effects, such as E(v)=0.

Models with different distributions and link functions

By assuming different distributions of y\mid u and u, and using different functions of g and 'v, we will be able to obtain different models. Moreover, the generalized linear mixed model (GLMM) is a special case of the hierarchical generalized linear model. In hierarchical generalized linear models, the distributions of random effect u do not necessarily follow normal distribution. If the distribution of u is normal and the link function of v is the identity function, then hierarchical generalized linear model is the same as GLMM.

Distributions of y\mid u and u can also be chosen to be conjugate, since nice properties hold and it is easier for computation and interpretation.

A summary of commonly used models are:

Fitting the hierarchical generalized linear models

Hierarchical generalized linear models are used when observations come from different clusters. There are two types of estimators: fixed effect estimators and random effect estimators, corresponding to parameters in : \eta = \mathbf{x}\boldsymbol\beta and in \mathbf{v(u)} , respectively. There are different ways to obtain parameter estimates for a hierarchical generalized linear model. If only fixed effect estimators are of interests, the population-averaged model can be used. If inference is focused on individuals, random effects will have to be predicted. There are different techniques to fit a hierarchical generalized linear model.

Examples and applications

Hierarchical generalized linear model have been used to solve different real-life problems.

Engineering

For example, this method was used to analyze semiconductor manufacturing, because interrelated processes form a complex hierarchy. Semiconductor fabrication is a complex process which requires different interrelated processes. Hierarchical generalized linear model, requiring clustered data, is able to deal with complicated process. Engineers can use this model to find out and analyze important subprocesses, and at the same time, evaluate the influences of these subprocesses on final performance.

Business

Market research problems can also be analyzed by using hierarchical generalized linear models. Researchers applied the model to consumers within countries in order to solve problems in nested data structure in international marketing research.

References

References

1. (1989). "Generalized Linear Models". Chapman and Hall/CRC.
2. Agresti, Alan. (2002). "Categorical Data Analysis". John Wiley & Sons, Inc..
3. (2009). "Identifiability of Parameters in Latent Structure Models with Many Observed Variables". The Annals of Statistics.
4. (Dec 2010). "hglm: A Package for Fitting Hierarchical Generalized Linear Models". The R Journal.
5. (2011). "Hierarchical Modeling Using Generalized Linear Models". Quality and Reliability Engineering International.
6. (2000). "A machine learning approach to yield management in semiconductor manufacturing". International Journal of Production Research.
7. (2011). "Interpreting interrelations across multiple levels in HGLM models: An application in international marketing research". International Marketing Review.