How to Create Non-linear Models

This is a quick guide to plotting curves using Tuva's functioning editor, f(x).

You can use the function editor to model the relationship between two quantitative attributes or variables.

The mechanics are simple—drag two quantitative attributes on the x and y axes, and fit a curve to approximate the relationship between the two attributes.

Step 1: Create the Scatterplot

Drag quantitative attributes to both the x and y axes.

Note: If you do not plot both an x and y attribute, the f(x) functionally will not allow you to plot a curve.

Step 2: Open the Modeling Card

To model the relationship (which in this context is between roller coaster height and speed), choose f(x) from the toolbar to open the modeling card. The modeling card will appear on the left-hand side of your screen.

In the modeling card, you can either enter the function in the editor, or choose the desired function form from the drop-down menu. If you choose a function from the drop-down menu, you may need to adjust the function for your own context.

When you have done this, two things happen:

1. Your function will appear on the graph. If it doesn’t, you will need to estimate the range of the parameter to get it in the viewing window, and/or zoom out. Click here for more detailed instructions.

2. The parameter (g in this case) appears as a slider below the function editor in the modeling card.

Step 4: Tweak the Parameter

You can change the value of a parameter by dragging the slider pointer.  As you do this, the curve will move.

The default lower bound of the slider is -10 and the default upper bound is +10. You can tweak the minimum and maximum values by clicking on them and inputting the desired value. Hit enter to submit the maximum and minimum values.

Alternatively, you can type in a new value for the parameter and press enter.

Additionally, if you feel that you need to make very small or large changes to the parameter value, double click on the upper range of the slider and adjust the steps to the desired value.

Try it Yourself!

Practice tweaking the parameter for this dataset, here. See if you can find a good fit. (Note: You will need to open the modeling card by clicking f(x) from toolbar, as shown above).

Step 5: Finding the Right Fit

Continue to tweak the parameter value until you get the closest possible fit. In this case, you get a reasonable fit between 9.0 and 9.5.

Once you are happy with your fit, you are all set to start making sense of your function in the context of the data!