Road Traffic Accident Classification Using Open Government Data

Generated outputs from a Random Forest classifier trained to predict road 
traffic accident severity (1=fatal, 2=serious, 3=slight) using the STATS19 
North Yorkshire dataset (2009–2013). Contains: test-set predictions (1,672 
rows with true/predicted severity labels and class probabilities), evaluation 
metrics (accuracy=0.751, weighted F1=0.673, macro F1=0.322, ROC-AUC=0.605), 
and a confusion matrix figure (300 dpi PNG). Model: RandomForestClassifier, 
100 estimators, balanced class weights, random_state=42. Data loaded from 
DBRepo REST API view ml_accident_features (8,358 total records, 6,686 train, 
1,672 test).