Огромное спасибо! Все отлично.
Информация о работе
Подробнее о работе
methods+of+eliminating+multicollinearity
- 15 страниц
- 2015 год
- 105 просмотров
- 0 покупок
Гарантия сервиса Автор24
Уникальность не ниже 50%
Фрагменты работ
Multicollinearity is a correlation of two or more explanatory variables in the regression equation. Estimation of coefficient regression equation may be insignificant, not only because of this factor for minor, but due to the fact that it is difficult to differentiate the impact on the final two or more of the variable factors. This usually occurs when the factors, which are linearly linked, changing synchronously. The nature of multicollinearity can be identified most clearly by the example of perfect multicollinearity, if the factors are functionally related to each other.
The aim of this work is a theoretical overview of methods to eliminate the multicollinearity.
Introduction 3
1 Multicollinearity: the concept of Major 4
2 Methods of elimination of multicollinearity 8
Conclusion 13
List of references 15
Thus, on the basis of the above, a number of conclusions.
The reason for multicollinearity is the property of an economic system in which all the variables of economic objects should be interconnected. Otherwise, the economic system will not exist. Therefore, multicollinearity was, is and will be, it is useless to fight, but to consider it in the construction of multifactor models is necessary.
Although reliable methods to identify collinearity doesn’t exist, there are several signs to identify it:
- Characteristic sign of multicollinearity is a high value of the coefficient of determination in the insignificance of the parameters of the equation (by t-statistics);
- In models with two variables is the best sign of multicollinearity correlation coefficient;
- In the model with a large number (less than two) the correlation coefficient factors may be low due to the presence multicollinearity should take into account the partial correlation coefficients;
• If the coefficient of determination is great, and partial factors are small, the possible multicollinearity.
Common approach to the elimination of multicollinearity exists, there does not exist. There are a number of methods that aren’t universal and are applicable to specific situations.
The simplest method to eliminate multicollinearity is an exception to the pattern of one or more correlated variables. It requires careful not to drop the variable that is required in the model in its economic essence, but it is often correlated with other variables (for example, the price of goods and prices of substitute goods).
Sometimes to eliminate the multicollinearity enough to increase the sample size. For example, using of annual data can go to the quarterly data. This will lead to a reduction in the variance of the regression coefficients and an increase in their significance. However, it is possible to strengthen the autocorrelation, which limits this approach.
In some cases, changes in model specification, for example, adding a significant factor, solve the problem of multicollinearity.
In some cases, you can minimize or completely eliminate the problem of multicollinearity by converting variables.
1. A teaching aid to study the course "Statistics". NN Schurenko, GV Devlikamiova: Ufa, 2004. - 55 p.
2. Econometrics for beginners. Basic concepts, basic techniques, the limits of applicability, interpretation of the results. Nosko VP: Moscow, 2000. - 249 p.
3. Yeliseyev II. Econometric: Moscow "Finance and Statistics", 2013. - 338 p.
Форма заказа новой работы
Не подошла эта работа?
Закажи новую работу, сделанную по твоим требованиям
