Linear regression¶

This method deals with the parametric modelling of a probability distribution for a random vector . It aims to measure a type of dependence (here a linear relation) which may exist between a component and other uncertain variables .

The principle of the multiple linear regression model is to find the function that links the variable to other variables ,…, by means of a linear model:

where describes a random variable with zero mean and standard deviation independent of the input variables . For given values of ,…,, the average forecast of is denoted by and is defined as:

The estimators for the regression coefficients , and the standard deviation are obtained from a sample of , that is a set of values ,…,. They are determined via the least-squares method:

In other words, the principle is to minimize the total quadratic distance between the observations and the linear forecast .

Some estimated coefficient may be close to zero, which may indicate that the variable does not bring valuable information to forecast . A classical statistical test to identify such situations is available: Fisher’s test. For each estimated coefficient , an important characteristic is the so-called “-value” of Fisher’s test. The coefficient is said to be “significant” if and only if is greater than a value chosen by the user (typically 5% or 10%). The higher the -value, the more significant the coefficient.

Another important characteristic of the adjusted linear model is the coefficient of determination . This quantity indicates the part of the variance of that is explained by the linear model:

where denotes the empirical mean of the sample .

Thus, . A value close to 1 indicates a good fit of the linear model, whereas a value close to 0 indicates that the linear model does not provide a relevant forecast. A statistical test allows to detect significant values of . Again, a -value is provided: the higher the -value, the more significant the coefficient of determination.

By definition, the multiple regression model is only relevant for linear relationships, as in the following simple example where .

In this second example (still in dimension 1), the linear model is not relevant because of the exponential shape of the relation. But a linear approach would be useful on the transformed problem . In other words, what is important is that the relationships between and the variables ,…, is linear with respect to the regression coefficients .

The value of is a good indication of the goodness-of fit of the linear model. However, several other verifications have to be carried out before concluding that the linear model is satisfactory. For instance, one has to pay attentions to the “residuals” of the regression:

A residual is thus equal to the difference between the observed value of and the average forecast provided by the linear model. A key-assumption for the robustness of the model is that the characteristics of the residuals do not depend on the value of : the mean value should be close to 0 and the standard deviation should be constant. Thus, plotting the residuals versus these variables can fruitful.

In the following example, the behavior of the residuals is satisfactory: no particular trend can be detected neither in the mean nor in he standard deviation.

The next example illustrates a less favorable situation: the mean value of the residuals seems to be close to 0 but the standard deviation tends to increase with . In such a situation, the linear model should be abandoned, or at least used very cautiously.

API:

Examples: