Main Content

The hat matrix provides a measure of leverage. It is useful for investigating whether one or more observations are outlying with regard to their *X* values, and therefore might be excessively influencing the regression results.

The hat matrix is also known as the *projection matrix* because it projects the vector of observations, y, onto the vector of predictions, $$\widehat{y}$$, thus putting the "hat" on y. The hat matrix *H* is defined in terms of the data matrix *X*:

*H* = *X*(*X ^{T}X*)

and determines the fitted or predicted values since

$$\widehat{y}=Hy=Xb.$$

The diagonal elements of *H*, *h*_{ii}, are called leverages and satisfy

$$\begin{array}{l}0\le {h}_{ii}\le 1\\ {\displaystyle \sum _{i=1}^{n}{h}_{ii}}=p,\end{array}$$

where *p* is the number of coefficients, and *n* is the number of observations (rows of *X*) in the regression model. `HatMatrix`

is an *n*-by-*n* matrix in the `Diagnostics`

table.

After obtaining a fitted model, say, `mdl`

, using `fitlm`

or `stepwiselm`

, you can:

Display the

`HatMatrix`

by indexing into the property using dot notationWhenmdl.Diagnostics.HatMatrix

*n*is large,`HatMatrix`

might be computationally expensive. In those cases, you can obtain the diagonal values directly, usingmdl.Diagnostics.Leverage

Leverage is a measure of the effect of a particular observation on the regression predictions due to the position of that observation in the space of the inputs. In general, the farther a point is from the center of the input space, the more leverage it has. Because the sum of the leverage values is *p*, an observation *i* can be considered as an outlier if its leverage substantially exceeds the mean leverage value, *p*/*n*, for example, a value larger than 2**p*/*n*.

The leverage of observation *i* is the value of the *i*th diagonal term, *h*_{ii}, of the hat matrix, *H*, where

*H* = *X*(*X*^{T}*X*)^{–1}*X*^{T}.

$$\begin{array}{l}0\le {h}_{ii}\le 1\\ {\displaystyle \sum _{i=1}^{n}{h}_{ii}}=p,\end{array}$$

where *p* is the number of coefficients in the regression model, and *n* is the number of observations. The minimum value of *h*_{ii} is 1/*n* for a model with a constant term. If the fitted model goes through the origin, then the minimum leverage value is 0 for an observation at *x* = 0.

It is possible to express the fitted values, $$\widehat{y}$$, by the observed values, *y*, since

$$\widehat{y}=Hy=Xb.$$

Hence, *h*_{ii} expresses how much the observation *y _{i}* has impact on $${\widehat{y}}_{i}$$. A large value of

`Leverage`

is an `Diagnostics`

table.After obtaining a fitted model, say, `mdl`

, using `fitlm`

or `stepwiselm`

, you can:

Display the

`Leverage`

vector by indexing into the property using dot notationmdl.Diagnostics.Leverage

Plot the leverage for the values fitted by your model using

See theplotDiagnostics(mdl)

`plotDiagnostics`

method of the`LinearModel`

class for details.

This example shows how to compute `Leverage`

values and assess high leverage observations. Load the sample data and define the response and independent variables.

```
load hospital
y = hospital.BloodPressure(:,1);
X = double(hospital(:,2:5));
```

Fit a linear regression model.

mdl = fitlm(X,y);

Plot the leverage values.

plotDiagnostics(mdl)

For this example, the recommended threshold value is 2*5/100 = 0.1. There is no indication of high leverage observations.

`LinearModel`

| `fitlm`

| `stepwiselm`

| `plotDiagnostics`