We aimed to forecast individual behaviors in a range of incident scenarios in our decision tree analysis. These included circumstances in which people did not flee and others in which they did so by vehicle, foot, or by some other means. With an accuracy of almost 67%, our model demonstrated a significant capacity for accurate prediction in the majority of instances. This degree of precision highlights the model’s overall efficacy in evaluating and deciphering a range of behavioral reactions across various incident settings.
On the other hand, more examination of the confusion matrix showed some incorrect classifications. The algorithm correctly predicted 676 fleeing episodes and successfully detected 37 cases in which people did not escape, but it also incorrectly classified a number of cases. Notably, in 125 cases, the model predicted fleeing when it didn’t actually happen.However, 33 cases were mistakenly predicted to be escaping on foot, and 136 cases were inaccurately labeled as being in a car. These incorrect classifications highlight particular parts of the model that need to be improved in order to increase its dependability and forecast accuracy.