Forecasting road traffic accident using deep artificial neural network approach in case of Oromia Special Zone

Millions of people are dying, and billions properties damaged by road traffic accidents each year worldwide. In the case our country Ethiopia, effect is even more causing injuries, death, property damage. Forecasting accident predicting severity contributes a role indirectly in reducing accidents. This study deals with forecasting number prediction an Oromia Special Zone using deep artificial neural network models. Around 6170 data collected from Police Commission Excel Traffic Department; dataset consists Districts (Woredas) 2005 to 2012 15 attributes. 5928 or (80%) was used for training model, 1482 (20%) testing model. proposed six different architectures such as backpropagation (BPNN), feedforward (FFNN), multilayer perceptron (MLPNN), recurrent networks (RNN), radial basis function (RBFNN) long short-term memory (LSTM) models LSTM model time serious within specified years. The will take input data, classify accidents, predict accident. Accident predictor GUI has been created Python Tkinter library easy prediction. According performance results, RNN showed best accuracy 97.18%, whereas MLP, LTSM, RBFNN, FFNN, BPNN 97.13%, 91.00%, 87.00%, 80.56%, 77.26%, respectively. LTSM forecasted three years which 3555 where actual 3561. forecast result obtained be helpful planning management