ANAKIN - Automated Analysis & Classification of Delay Causes in the Railway System

Systematic monitoring of delays is carried out continuously in the field of rail traffic. Measuring train delays, in relation to the timetable, is a relatively simple task that is currently automated in Sweden. Identifying the root cause of a particular delay, as well as deriving how the delay might propagate in the system and between trains, is more difficult.

Purpose of the project

This project aims to analyze the Swedish Transport Administration's manual process for cause coding of delays and investigate whether there are conditions to facilitate both the initial cause coding and the review of this through appropriate support functions.

Implementation of the project

As the project includes mapping of processes, assessment and decisions regarding cause coding and development of decision support technology, complementary qualitative and quantitative approaches are required.

In the project's initial two work packages, a review of the current state of knowledge is made in the form of a literature review of relevant reports and research. The literature review is complemented by empirical data collection, which focuses on organizational and human aspects of causal coding assessment and decision-making.

In work package 3, we focus on how automated machine learning methods could be used in a decision support role in the organization.

Within the framework of this project, initial evaluations will be carried out for e.g. the derived model using relevant evaluation metrics. This will give an idea of how well the derived model can be considered to work on existing input data.