Proofs involving ordinary least squares


(Learn how and when to remove this template message)

The purpose of this page is to provide supplementary materials for the ordinary least squares article, reducing the load of the main article with mathematics and improving its accessibility, while at the same time retaining the completeness of exposition.

Derivation of the normal equations edit

Define the  th residual to be


Then the objective   can be rewritten


Given that S is convex, it is minimized when its gradient vector is zero (This follows by definition: if the gradient vector is not zero, there is a direction in which we can move to minimize it further – see maxima and minima.) The elements of the gradient vector are the partial derivatives of S with respect to the parameters:


The derivatives are


Substitution of the expressions for the residuals and the derivatives into the gradient equations gives


Thus if   minimizes S, we have


Upon rearrangement, we obtain the normal equations:


The normal equations are written in matrix notation as

  (where XT is the matrix transpose of X).

The solution of the normal equations yields the vector   of the optimal parameter values.

Derivation directly in terms of matrices edit

The normal equations can be derived directly from a matrix representation of the problem as follows. The objective is to minimize


Here   has the dimension 1x1 (the number of columns of  ), so it is a scalar and equal to its own transpose, hence   and the quantity to minimize becomes


Differentiating this with respect to   and equating to zero to satisfy the first-order conditions gives


which is equivalent to the above-given normal equations. A sufficient condition for satisfaction of the second-order conditions for a minimum is that   have full column rank, in which case   is positive definite.

Derivation without calculus edit

When   is positive definite, the formula for the minimizing value of   can be derived without the use of derivatives. The quantity


can be written as


where   depends only on   and  , and   is the inner product defined by


It follows that   is equal to


and therefore minimized exactly when


Generalization for complex equations edit

In general, the coefficients of the matrices   and   can be complex. By using a Hermitian transpose instead of a simple transpose, it is possible to find a vector   which minimizes  , just as for the real matrix case. In order to get the normal equations we follow a similar path as in previous derivations:


where   stands for Hermitian transpose.

We should now take derivatives of   with respect to each of the coefficients  , but first we separate real and imaginary parts to deal with the conjugate factors in above expression. For the   we have


and the derivatives change into


After rewriting   in the summation form and writing   explicitly, we can calculate both partial derivatives with result:


which, after adding it together and comparing to zero (minimization condition for  ) yields


In matrix form:


Least squares estimator for β edit

Using matrix notation, the sum of squared residuals is given by


Since this is a quadratic expression, the vector which gives the global minimum may be found via matrix calculus by differentiating with respect to the vector   (using denominator layout) and setting equal to zero:


By assumption matrix X has full column rank, and therefore XTX is invertible and the least squares estimator for β is given by


Unbiasedness and variance of edit

Plug y =  + ε into the formula for   and then use the law of total expectation:


where E[ε|X] = 0 by assumptions of the model. Since the expected value of   equals the parameter it estimates,  , it is an unbiased estimator of  .

For the variance, let the covariance matrix of   be   (where   is the identity   matrix), and let X be a known constant. Then,


where we used the fact that   is just an affine transformation of   by the matrix  .

For a simple linear regression model, where   (  is the y-intercept and   is the slope), one obtains


Expected value and biasedness of edit

First we will plug in the expression for y into the estimator, and use the fact that X'M = MX = 0 (matrix M projects onto the space orthogonal to X):


Now we can recognize ε as a 1×1 matrix, such matrix is equal to its own trace. This is useful because by properties of trace operator, tr(AB) = tr(BA), and we can use this to separate disturbance ε from matrix M which is a function of regressors X:


Using the Law of iterated expectation this can be written as


Recall that M = I − P where P is the projection onto linear space spanned by columns of matrix X. By properties of a projection matrix, it has p = rank(X) eigenvalues equal to 1, and all other eigenvalues are equal to 0. Trace of a matrix is equal to the sum of its characteristic values, thus tr(P) = p, and tr(M) = n − p. Therefore,


Since the expected value of   does not equal the parameter it estimates,  , it is a biased estimator of  . Note in the later section “Maximum likelihood” we show that under the additional assumption that errors are distributed normally, the estimator   is proportional to a chi-squared distribution with n – p degrees of freedom, from which the formula for expected value would immediately follow. However the result we have shown in this section is valid regardless of the distribution of the errors, and thus has importance on its own.

Consistency and asymptotic normality of edit

Estimator   can be written as


We can use the law of large numbers to establish that


By Slutsky's theorem and continuous mapping theorem these results can be combined to establish consistency of estimator  :


The central limit theorem tells us that


Applying Slutsky's theorem again we'll have


Maximum likelihood approach edit

Maximum likelihood estimation is a generic technique for estimating the unknown parameters in a statistical model by constructing a log-likelihood function corresponding to the joint distribution of the data, then maximizing this function over all possible parameter values. In order to apply this method, we have to make an assumption about the distribution of y given X so that the log-likelihood function can be constructed. The connection of maximum likelihood estimation to OLS arises when this distribution is modeled as a multivariate normal.

Specifically, assume that the errors ε have multivariate normal distribution with mean 0 and variance matrix σ2I. Then the distribution of y conditionally on X is


and the log-likelihood function of the data will be


Differentiating this expression with respect to β and σ2 we'll find the ML estimates of these parameters:


We can check that this is indeed a maximum by looking at the Hessian matrix of the log-likelihood function.

Finite-sample distribution edit

Since we have assumed in this section that the distribution of error terms is known to be normal, it becomes possible to derive the explicit expressions for the distributions of estimators   and  :


so that by the affine transformation properties of multivariate normal distribution


Similarly the distribution of   follows from


where   is the symmetric projection matrix onto subspace orthogonal to X, and thus MX = XM = 0. We have argued before that this matrix rank n – p, and thus by properties of chi-squared distribution,


Moreover, the estimators   and   turn out to be independent (conditional on X), a fact which is fundamental for construction of the classical t- and F-tests. The independence can be easily seen from following: the estimator   represents coefficients of vector decomposition of   by the basis of columns of X, as such   is a function of . At the same time, the estimator   is a norm of vector divided by n, and thus this estimator is a function of . Now, random variables (, ) are jointly normal as a linear transformation of ε, and they are also uncorrelated because PM = 0. By properties of multivariate normal distribution, this means that and are independent, and therefore estimators   and   will be independent as well.

Derivation of simple linear regression estimators edit

We look for   and   that minimize the sum of squared errors (SSE):


To find a minimum take partial derivatives with respect to   and  


Before taking partial derivative with respect to  , substitute the previous result for  


Now, take the derivative with respect to  :


And finally substitute   to determine