Добавил:
Upload Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:
STATE_EXAM_ANSWERS_2012.doc
Скачиваний:
4
Добавлен:
01.05.2025
Размер:
2 Мб
Скачать
☆

25,26,27,29 Simple regression analysis. The Simple Linear Model. Least Squares Regression. Interpretation of a Regression Equation.

Regression analysis – describes and evaluates the relationship between a given (dependent) variable and other independent variables.

P – real economic data. X1 … four hypothetical values of the explanatory variables. If the relationship between Y and X were known the corresponding values of Y will be represented by the points Q1, Q2 … on the line. U - the disturbance term. U is positive in P1 and P4, and negative in P2 and P3. If you plot a line through the actual value of Y and X the line will cross P.

The actual values of B1 and B2 and hence the location of the Q points are unknown as are the values of the U in the observations. The task of the regression analysis is to obtain estimates of B1 and B2 (slopes) and hence to estimate the location of the line given by P points. B1, B2 are the regression coefficients.

Why does U exist? 1) Omission of explanatory variables due to impossibility of obtaining them. 2)Aggregation of the variables. 3)Model misspecification. The model may be misspecified in terms of its structure. 4)Functional misspecification. The relationship between Y and X may be misspecified mathematically. 4)Measurement error. If the measurement of one or more variables in the relationship is wrong the relationship between Y and X will be wrong too.

The equation of the fitted (regression) line is =B1+B2*X1

The caret mark over means that Is the fitted (подобранное) value of Y that corresponds to X, NOT the actual value. In figure – Q are the fitted points. B1 and B2 are only estimates and can be good or bad.

Calculating U. U is the difference between the actual and the estimated (fitted) value, and the distance between P1 and Q1. U is “e”.

Substituting we get

The e depends on the choice of B1 and B2 and should be close to zero. To solve this problem we should calculate RSS (residual Sum of Squares). If we can reduce RSS to zero we will have a perfect fit.

simple linear regression fits a straight line through the set of P points in such a way that makes the sum of squared residuals of the model (that is, vertical distances between the P and the fitted line (Q)) as small as possible. We minimize RSS by using Ordinary Least Squares analysis.

We can influence the size of RSS only by changing B1 and B2, while Y and X are fixed.

R2. For ex. R2=0.73 means that variance (дисперсия) of Y is described by this model on 73%. в числителе находим среднее значение ошибки и делим ее на дисперсию.

TSS=RSS+ESS

T SS(total sum of squares) is the sum, of the squares of the deviations (over all observations) of each observation from the overall mean;

ESS (explained sum of squares) is the sum of the squares of the deviations of the predicted values from the mean value of a response variable.

F-test (Fisher’s distribution)

N number of observations (степени свободы, количество X), K-number of parameters (X, Y).

It should be that F-test>Fcritic, means that the variance of “e” is equal. (дисперсия случайных величин, квадраты которых суммируются, одинакова.) and R2 is not random.

Goodness of fit describes how well it fits a set of observations. Shows the discrepancy between observed values and the values expected under the model in question. Such measures can be used in statistical hypothesis testing, e.g. to test for normality of residuals, to test whether two samples are drawn from identical distributions

Соседние файлы в предмете [НЕСОРТИРОВАННОЕ]