Regression Modeling
Tariq Hama Karim
Abstract- Data analysis is one of the most important elements of research methodology. It encompasses different classic and modern types such as descriptive, dispersion, factor, discriminant, fuzzy logic, evolutionary algorithms, time series and regression analysis. Most of regression models are empirical. These models are describing the functional relationship between a response variable and one or more predictor variables. Regression analysis used for forecasting, time series modeling and finding the causal- effect relationship between the variables.
The history of this particular statistical technique can be traced back to the work of Galton who showed that the height of individuals can be predicted from the height of their parents.
There are numerous types of regression models that you can use. This choice often depends on the kind of data you have for the dependent variable and the type of model that provides the best fit. The simplest forms are simple linear and multiple linear regression techniques. These types of models can be established when several assumptions (Normality, homoscedasticity, no serial correlation and no inter-correlation) are satisfied. When any of these assumptions are violated, sophisticated regression techniques can be applied. There are hundreds of techniques, each has its own significance. For instance PCR and Ridge regression deal with the problem of multicolinearity. These techniques are like medicines for different types of diseases. To avoid the problem of overfitting, one should be parsimonious, i.e., the least number of parameters should be used. Each model can be evaluated by cross validation or partitioning the data to training and test sets.
Keywords- Regression , Data analysis, normality, fitting model
Date: 15/11/2017
Place: VIP Hall