POINT OF INTEREST AND STOP TYPE PREDICTION FOR PARATRANSIT TRANSPORTATION DATAIN STUDY CASE : BANDUNG CITY ALGORITHM : ANN, K-MEANS

• Main Author: Permana Putri, Elbananda
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/55574
• Institution: Institut Teknologi Bandung

Paratransit vehicles often do not have a static stop point. Sometimes it causes a traffic jam, especially in Southeast Asia. In Indonesia, Angkot is a well-known paratransit vehicle. Angkot often causes congestion because it can stop anywhere and anytime. It is quite difficult to predict when and where the Angkot stops. This research proposes a machine learning model design that can predict the types of Angkot stops, as well as its Point of Interest. This research using Angkot trips history data in 2018. Data processing is carried out in several stages, consisting of manual labeling with google maps and google road maps as a reference, grouping data, increasing the correlation matrix to the output data, normalizing GPS data, finally entering the data into the Multi-Layer Perceptron model for training. A model that has been trained can labelling raw datawhich has the same route as the data being trained. The model also shows 95% accuracy with GPS data that has been normalized and clustered first.