First of all, i'm a giant fan of regression analyses; i exploit them on a daily. Its blessings and drawbacks rely upon the particular form of multivariate analysis that's conducted.
There therefore seems to be some ambiguity within the question, however this could be resolved easily: multivariate analysis doesn't refer entirely to (linear) (multiple) OLS-type of models. The term rather refers to a broad category of models, during which one or many outcome variables square measure sculptural via some regression operate, which will or might not be linear. however after all in daily apply, we frequently simply mean (linear) OLS regression.
Many of the (perceived) disadvantages of multivariate analysis normally square measure very specific issues of linear OLS models. These issues will promptly be resolved by applying a unique form of multivariate analysis. Non-linearity is also resolved via acceptable non-linear terms; heteroskedasticity via sturdy estimators; agglomeration via construction techniques, etc.
Regardless of the sort of regression, such associate degree analysis will go seriously wrong if there square measure severe outliers or powerful cases. These should be known and restricted consequently.
As the noted statistician Francis Anscombe showed along with his ill-famed “quartet” data: you'll be able to have ‘evidently different’ data-sets that yield identical regression model. The Anscombe quartet is already alright delineate within the topic: what's the importance of Anscombe's quartet?
Advantages of Regression analysis:
Regression analysis refers to a way of mathematically searching for that variables could have a control. The importance of multivariate analysis for atiny low business is that it helps verify that factors matter most, that it will ignore, and the way those factors move with one another. The importance of multivariate analysis lies within the undeniable fact that it provides a robust statistical procedure that permits a business to look at the connection between 2 or a lot of variables of interest.
Despite the on top of blessings, the technique of multivariate analysis suffers kind the subsequent serious limitations:
It is assumed that the cause and impact relationship between the variables remains unchanged. This assumption might not perpetually hold sensible and thence estimation of the values of a variable created on the premise of the equation could result in inaccurate and dishonest results.
The practical relationship that's established between any 2 or a lot of variables on the premise of some restricted information might not hold sensible if a lot of and a lot of information square measure taken into thought. for instance, just in case of the Law of come back, the law of decreasing come back could return to play, if an excessive amount of of inputs square measure used with ca read to increasing the degree of output.
It involves terribly protracted and sophisticated procedure of calculations and analysis.
It can not be utilized in case of qualitative development viz. honesty, crime etc.
Post a Comment for "Advantages and Disadvantages of Regression Analysis"