Here is the synopsis of our sample research paper on The Problems of Multicollinearity, Heteroskedasticity, Outlying and Influential Cases and Non-normally Distributed Errors When using Ordinary Least Squares (OLS) Regression. Have the paper e-mailed to you 24/7/365.
Essay / Research Paper Abstract
Multicollinearity, heteroskedasticity, outlying and influential cases, and non-normally distributed errors all present problems for ordinary least squares (OLS) regression. This 8 pager paper explains why each is a problem for OLS, how it can be detected and looks at steps can be taken to deal with each of them. The bibliography cites 5 sources.
Page Count:
8 pages (~225 words per page)
File: TS14_TEOSLerror.rtf
Buy This Term Paper »
Unformatted sample text from the term paper:
of data parameters and to fit data into a model is the least squares method. This is an old technique that the French mathematician Legendre first published in 1805,
but the method is older than this, as a later published memoir published in 1808 indicated that Gauss had used this approach in 1795 (Bates and Watts, 1988). The ordinary
least squares (OLS) has many advantages, it is easy to use and the basics are well understand, leading to a simple implementation. However, the model is not perfect and despite
is long history and wide use there are several potential errors that can arise and created flaws in the results. Multicollinearity, heteroskedasticity, outlying and influential cases, and non-normally distributed errors
all present problems for ordinary least squares (OLS) regression. To consider what these problems are we first need to examines the basis of the OLS model itself otherwise the way
the error may occur from these different potential phenomena cannot be fully appreciated. The main use of OLS is linear regression. This is the use of data where
it is assumed there is a correlation between the different sets of data which when plotted with give the plotter either a line or a curve that indicates the best
fit. In effect this is the line that gives the least squares on the graph between the line and the points that are plotted (Bates and Watts, 1988).
The standard formula has a set of N pairs, these observations {YiXi} is utilised in order to find the function that will give the value of Y, which is the
dependant variable resulting from X which is the independent variable (Bates and Watts, 1988). If there is one variable and there is a linear function the following equation will result
...