Adjust the youth profile to see the projected substance-use trajectory, the estimated effect of justice-system disposition, a risk estimate, and the model's own explanation of that estimate. Everything updates live as you change the profile.
–
1
Trajectory subgroup
trajectory model
–
this youth's projected classlow/stable reference
2
Effect of disposition
causal estimate
lowers outcomeraises outcome
3
Risk estimate
risk model
–
0%25%50%75%100%
Top contributing factors
4
Summary
combined
5
Why this estimate
explainable AI
Local explanation
feature attribution · this youth
Bar lengths are additive in log-odds; the ends show the resulting probability. Negative bars lower the estimate, positive bars raise it.
Feature shape
partial dependence · entry age
Risk vs entry age with all other factors held at this youth's values. The entry-age effect is modeled below age 15 and is flat above it.
Global importance
mean |attribution| (log-odds)
How much each factor typically moves the estimate across profiles, independent of this youth.