This Blog entry is from the Linear Regression section in Learn Palisade.
A one-way linear regression model is the first model created in this Blog that has predictive power. In actuality, a linear regression model was brought to bear in drawing a trend line, however this Blog entry introduces a more detailed means to create a predictive model via linear regression.
Having ranked the correlation strength of the independent variables vs the dependent variable, note the strongest correlating value which is in this case the variable Kurtosis. The strongest correlating variable will always be the starting point in the development of a predictive model in this methodology.
For the dataset, in the StatTools ribbon, select the menu Regression and Classification towards the centre right of the ribbon, then click on the sub menu Regression (referring to linear regression):
The regression window will appear, shifting from the nomenclature of x and Y (referring to the axis on a chart) to Dependent and Independent (D and I respectively):
The dependent variable for the purpose of these Blog entries is intuitively labelled Dependent in the datasets. For the variable Dependent, ensure that the D column is selected (only one can be selected):
The Independent Variable, being the first and strongest correlating value in this example Kurtosis will be selected in the I column I:
The default parameters are adequate for the model, thus clicking OK will set about calculating a Linear Regression model. StatTools will return a linear regression output: