Fig. 5From: Quantifying long-term impact of court decisionsThe empirical distribution of the average relative error ε of the model against the data (ensembled over all decisions with more than 30 citations) decays exponentially with increasing error, indicating that the model describes the majority of the cases. The relative error of a decision i is defined as \(\epsilon _{i}\equiv \frac {1}{t^{\max }}\sum _{t=0}^{t^{\max }}\left |x_{i}(t)-x(t)\right |/x_{i}(t)\) where x(t) corresponds to Eq. 2 with the best parameters obtained from MLEBack to article page