Time aggregation & relative comparison (e.g. YoY)
In charts and tables, and in the roll-up columns in a spreadsheet, Causal can automatically aggregate a variable's values across time. For example: if your model is monthly and you have a Revenue variable, then Causal can automatically summarize the total revenue in each year.
The Time Aggregation function lets you choose how this summary is calculated. You can change the method by right-clicking on a variable and selecting Aggregation Function from the context menu -> Time Aggregation, or if you have Rollup columns toggled on you can click on a rollup cell to easily edit the Time Aggregation setting for each variable.
time aggregation in the right-click menu of a variable
time aggregation in the roll-up column cells
There are a few options:
Sum: this sums up the individual values. Examples:
- Anything that represents a discrete quantity which can accumulate, e.g. "Revenue"
Average: this calculates the mean of the individual values. Examples:
- Generally anything that's a percentage, like "Growth rate"
Final: this takes the final value in the set of individual values. Examples:
- Anything representing an accumulation, like "Total users" or "Cash in bank"
- A percentage that you expect to move roughly monotonically
Initial: just like final, but it takes takes the first value in the set of individual values.
Formula: uses the variable's formula to calculate an aggregation value. For example if you have a variable for the margin with the formula profit / revenue, the formula aggregation will take the total profit and divide it by the total revenue of the aggregated period.
Note that the Time Aggregation setting of a variable also determines how a variable is 'rolled up' to a model with a higher-level granularity. For example if you have a Weekly model, and then a Monthly model pulls in a variable from the Weekly model, should it pull the sum, the average, the final...?
You can easily compare the value of a variable across relative time periods (e.g. the prior month, the prior year). The comparison is shown on a time aggregated basis too, if relevant. For example, if your model is monthly and you've set your table to Yearly time aggregation, a YoY comparison will show for 2022 vs. 2021, as well as each month (e.g. Jan 2022 vs. Jan 2021).