Who do you treat, and what does it buy?

A budget covers a fraction of the population, so somebody chooses the fraction. Move the slider and watch both answers change: who is in the treated set, and how much outcome the budget actually buys.

Every number here was computed by itx demo build from the same benchmark that produces the repository's results table. Nothing is modelled in the browser, and no feature values are shipped: units are row numbers in a public dataset's test split.

These are one split, not the headline. The results table averages five random splits of each dataset; this page shows the first of them, because a slider has to move through a single ranking of actual units rather than an average of five different ones. So the numbers here sit near the table's and will not equal them. On ACIC at a 10% budget, for instance, this split gives the risk-ranking baseline -0.2434 where the five-split average is -0.1971. The repository carries the averages and their intervals.