Scatter plots | Vose Software

Scatter plots

 

Scatter plots show a dot for a number of samples that were generated during the run of the model, plotting the values of two variables against each other for each sample. ModelRisk will allow scatter plots between any two variables, irrespective of whether they are inputs or outputs. They are used to visualize the relationship between variables.

The following is an example of a simple scatter plot:

The horizontal axis shows the value of the Market growth, and the vertical axis shows the corresponding value of the NPV. The plot shows a positive correlation, i.e. if the Market growth is larger, the NPV will be larger (on average) too. We can also see, for example, that if the Market growth was 10%, the NPV would be between $20M and $32M or thereabouts.

The plot was created with 1000 values, the default, which gives a good impression of the relationship. Too many points, and this can become so dense one cannot see where the more likely scenarios lie:

Too few and the pattern is not very clear:

In general we recommend between 1000 and 3000 points.

There are two versions of the scatter plot: actual values of the variables, and the cumulative percentiles. The following plot shows the cumulative percentiles with the Slider option on:

The axes are split here into equal partitions (thirds). If there was no correlation, a roughly equal fraction of points would lie in the nine squares, which clearly isn’t the case and confirms statistically that a correlation exists. The horizontal and vertical sliders can be repositioned.

The format of the plot can be comprehensively adjusted by clicking on the various control buttons in the Scatter Options tab:

The following editing tabs are available:

 

Navigation