COMBINED METHOD OF LSTM AND RANDOM FOREST FOR TRAFFIC JAM PREDICTION

• Main Author: Arfina, Ayuni
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/85301
• Institution: Institut Teknologi Bandung

Traffic jam is a crucial issue in several cities in Indonesia, including cities in West Java province. Traffic jams cause many losses such as lost time, lack of productivity, and increased stress for the driver, and that will have an impact on the next activity of the day. Traffic jams are caused by the growth of vehicles, which has increased quite significantly and is not comparable to the growth of highways, and traffic jams are also caused by trips that are not well planned. This study proposes a method to create a traffic jam prediction model, so drivers can plan their trips well and reduce traffic jams. The method proposed in this study is a combined LSTM and Random Forest algorithm. LSTM is a deep learning algorithm that is well known for dealing with temporal and sequential issues, while Random Forest is a machine learning algorithm that has been widely used by previous researchers to predict traffic jams and provide excellent results. Random Forest has the advantages of being strong and stable, as well as having a low risk of overfitting. Many related studies have been carried out before, but no one has conducted experiments using a combination of the LSTM and Random Forest algorithms. The results of this study succeeded in proving that the combined LSTM and Random Forest algorithms give better results than the LSTM and Random Forest models that stand alone for this case. The combined model gives results of accuracy 28% better than Random Forest which stands alone, and 1% better than the LSTM model which stands alone.