Regression Modeling and diagnostics

The previous article was about regression generation data, for that we are now doing regression modeling and diagnostics.

When faced with regression analysis, we all know that in this analysis model estimation, testing, and determination of confidence intervals for the regression coefficients and model estimators are carried out. To obtain a feasible model, it is also necessary to explore the data prior to model estimation and diagnostics to model estimators (Mattjik, 2013). In this several methods useful for diagnosing violations of the basic regression assumptions. These diagnostic methods are primarily based on study of the model residuals. Methods for dealing with model inadequacies, as well as additional, more sophisticated diagnostics.

In simple terms, the residuals is defined as the difference between the data and the estimate.

The following information is obtained from the residuals:

1. Can see the distribution pattern of the random variable Y,

2. Through the residuals, we can find out whether the assumptions requirements in the estimation with MKT are met or no;

3. Through the residuals, we can also test the regression parameters, so we need to know the distribution of the residuals;

4. Through the residuals, we can also see if the model we choose fit or not;

5. Through residuals, we can also see if an observation is an outlier or no;

6. Through residuals, we can also see if an observation is an influential observation or not.

In this, more emphasis is placed on the visualization aspect in residual diagnostics, so that it is clear how the image (shape) will look if there is a violation of the rule. Although this technique is actually very rarely done by beginners. but with a technique like this we can get to know more in the form of violations of an assumption. The code or R programming language can be seen in my writing at this address https://rpubs.com/Muayyad/799151 and https://rpubs.com/Muayyad/779130. Thank U :))