Having set up the model, we can now inspect the data to determine if it is appropriate for modelling.

On the Model pane, click the Refresh button.

This action will fetch appropriate data for both the dependent and independent points from the database and display as a scatter graph.

Since the data in this scatter is not too widely spread, it would seem to imply that there is a relationship between the two points and is therefore appropriate for regression modelling. In a case where the data was widely spread then there is no point in proceeding further and the model parameters should be adjusted to try and create a data set that is more appropriate. If none can be found, then the model should be deleted.

When we are satisfied that we have a valid data set for modelling, we can run the model through the ‘R’ engine. To do this, simply press the ‘R’ button.

This will produce the following:

Our regression type is currently set to ‘Single’. We therefore get a single regression line. In addition, we have two further ‘confidence’ lines.

If we now expand the two rows in the ‘Model Outputs’ pane, we can see the numeric results from the model.

Here we can see a number of statistical parameters plus the gradient and intercept of the regression line.

Other Regression Types

We can now look at the other regression types. We can change it to ‘Segmented’ and press the ‘R’ button again to re run the model.

This is the chart displayed:

We now see that there are two regression lines indicating that the relationship changes around x=60.

If we now look at the Model Outputs again we see:

We can see that most of the parameters are the same, since we are using the same data set, however, we now have two regression line rows. One for each of the regression line segments.

The third regression type is ‘Polynomial’.

With this set and re-running the model once more we get:

This time we have curves which represent the regression and confidence lines.

Looking at the Model Outputs again we see:

Again the majority of parameters are the same, however we now have no regression line parameters since we now have no straight lines