Linear logistic regression model

Figure 3. The logistic regression model used to estimate LD50 is represented on the left where 7C represents the proportion of dead plants. The logistic curve can be linearized by using the logit transformation shown on the right. LD50 values were estimated with the regression coefficients for logit 7C=0.0, as shown in the inset box.

In the near future, more sophisticated models can be built using probabilistic networks. A probabilistic network is a factorization of the joint probabiHty function over all the considered variables (markers, interventions, and outcomes) based on knowledge about the dependencies and independencies between the variables. Such knowledge is naturally provided by the hits coming out of the association screen, where each association can be interpreted as a dependency, and the absence of an association as an independency between variables. The model can then be parameterized by fitting to the data, similarly to the linear and logistic regression models, which are in fact special cases of probabilistic network models. 

In Chapter 8 we show how to do computational Bayesian inference on the logistic regression model. Here we have independent observations from the binomial distribution where each observation has its own probability of success. We want to relate the probability of success for an observation to known values of the predictor variables taken for that observation. Probability is always between 0 and 1, and a linear function of predictor variables will take on all possible values from -oo to oo. Thus we need a link function of the probability of success so that it covers the same range as the linear function of the predictors. The logarithm of the odds ratio... 

Nelder and Wedderburn (1972) extended the general linear model in two ways. First, they relaxed the assumption that the observations have the normal distribution to allow the observations to come from some one-dimensional exponential family, not necessarily normal. Second, instead of requiring the mean of the observations to equal a linear function of the predictor, they allowed a function of the mean to be linked to (set equal to) the linear predictor. They named this the generalized linear model and called the function set equal to the linear predictor the link function. The logistic regression model satisfies the assumptions of the generalized linear model. They are ... 

In the logistic regression model, the observations are independent binomial observations where each observation has its own probability of success that is related to a set of predictor variables by setting the logarithm of the odds ratio equal to an unknown linear function of the predictor variables. This is known as the logit link function. The coefficients of the linear predictor are the unknown parameters that we are interested in. 

The logistic regression model is an example of a generalized linear model. The observations come from a member of one-dimensional exponential family, in this case binomial. Each observation has its own parameter value that is linked to the linear predictor by a link function, in this case the logit link. The observations are all independent. 

This classification of bonds allowed the application of logistic regression analysis (LoRA), which proved of particular benefit for arriving at a function quantifying chemical reactivity. In this method, the binary classification (breakable or non-breakable, represented by 1/0, respectively) is taken as an initial probability P0, which is modelled by the following functional dependence (Eqs. 7 and 8) where f is a linear function, and x. are the parameters considered to be relevant to the problem. The coefficients c. are determined to maximize the fit of the calculated probability of breaking (P) as closely as possible to the initial classification (P0). 

Physico-chemical measurements using chromatographic methods produce responses that are linear to the concentrations. As IA measures the resulting signals of a reaction, however, the response is a nonlinear function of the analyte concentration. Often, the regression model used to describe this relationship is a four- or five-parameter logistic function, as shown in the sigmoid shape standard curve in Fig. 6.4. 

In other words, what is a main effect in logistic regression terms becomes an interaction in log-linear model terms. However, it is really the log-linear model that is the odd one out among generalized linear models as regards use of interactions and, in more conventional terms, Simpson s paradox does not involve interactions. 

For a continuous child node both with discrete and continuous parent nodes (hybrid nodes), we can use, for example, a mixture (conditional) Gaussian distribution with the population mean of a child node as the linear combination of the continuous parent conditional distributions across different states of the discrete parent node. Conversely, for a discrete child node with hybrid nodes, we can apply, for example, logistic regression or probit models for inferring conditional distributions of the discrete child node variable. For example, in a logistic regression framework, we use... 

Figure 6 Plot of fluorescence observations from LightCycler (Roche Diagnostics). Forty observations give a sigmoid trajectory that can be described by full data fit (four-parametric logistic model). Ground phase can be linearly well regressed. FDM and SDM denote position of FDM and SDM within full data fit.

---