I have two different mathematical models I use to predict the dollar amount in claims a group will incur over the course of a year. I want to visually juxtapose how accurate those two models are on historical data (i.e. I know what actual claims for each group are) with group-level granularity. I haven't done much with data visualization in the past, so I was wondering if there are any pre-existing data-visualization paradigms that this kind of thing might fit into. Thanks.
In general before approaching any visualization you can use these links to understand more about what kind of data visualization you would need -
Now, as you have a date field (year), a numerical field (claims) which you want to show for different groups (categorical data) and you have output of two models and a actual values. ( so 3 different sources) .
Considering this you can create a line chart with time based data on x axis and claims on the y axis. The claims value can be shown based on the sources(historical data, model 1 and model 2) for comparison and you can provide a filter which will serve as a filter for groups.
This concept I just shared can be easily done in Tableau and you can see yourself.
I think a line graph is fine (as @cap suggested). To make things less noisy and more obvious, though, I suspect you may want to take the difference of your predicted value and your actual value; if you plot that on a log scale, (effectively plotting the log of error over time) you should get a really good understanding of the temporal aspects of the errors of your model.