Predictive Modelling using Machine Learning Techniques for Railway incident based parametersDenmark Railways

With the growing concerns for global warming and climate change immense efforts are being taken to reverse the situation. About 15% of the carbon dioxide contribution comes from the transportation sector, making the transportation sector one of the major contributor to global warming. Among the various modes of transport, railways are considered to be the most sustainable among all. But due to its unreliability and lack of punctuality it is increasingly becoming the less preferred mode than the others as they provide more flexibility and control. To make railways more attractive one of the ways is it make them more reliable by making its Traffic Management Systems more robust. Railway Traffic Management Systems(RTMS) ensure the smooth running of the operations. One of the major tasks of conventional RTMS is to reduce train delays. This becomes a very microscopic goal and leaves out the big picture of the situation on the network. This is the initial motivation to investigate and use Total Delay caused by an incident and the Total Duration it lasts, as the trigger for a TMS to initiate rescheduling rather than delays in trains itself. In this thesis we investigate this notion by trying to create a model for predicting the two mentioned attributes by using the real incident data from the country of Denmark. We explore the possible attributes that could affect total delay and the duration and using these attributes, creation of a predictive model is attempted. Standalone Neural Network models are created for both Total Delay and Duration. Other models XGBOOST, Generalised Linear Models, Linear Models were also created for the comparison with the neural networks and based on the metric results from all the models the best models for both the response variables are chosen. The chosen models are used to create the operational framework. This process could help in changing the way the conventional rescheduling systems work by considering parameters discussed here than the conventional parameters.      «

With the growing concerns for global warming and climate change immense efforts are being taken to reverse the situation. About 15% of the carbon dioxide contribution comes from the transportation sector, making the transportation sector one of the major contributor to global warming. Among the various modes of transport, railways are considered to be the most sustainable among all. But due to its unreliability and lack of punctuality it is increasingly becoming the less preferred mode than the...     »