Perm Polytechnic Develops Neural Network to Assess Urban Transit Accessibility
The tool could help planners design new developments.
Researchers at Perm Polytechnic have developed a neural network model that predicts transit accessibility in urban neighborhoods. Rather than simply generating estimates, the algorithm explains how conditions could change in the future.
“To create the model, researchers analyzed a real Russian city with different types of development – historic, Soviet-era and modern. They divided it into neighborhoods and collected planning data for each one, including street density, the number of intersections and dead ends, and how winding the streets are,” the university’s press service said.
The researchers paid particular attention to access to jobs. They ultimately developed a dedicated index that measures the number of workplaces and other destinations people can reach within 45 minutes.
“Knowing the current level of accessibility is no longer enough – traditional methods can already do that. What is needed is a tool that can assess the impact of planning changes in advance. Using two datasets – neighborhood planning parameters and their corresponding indices – we created a neural network model. It can predict transit accessibility and show which changes to street layouts will affect how convenient trips are. This will make it possible to analyze both areas that have yet to be built and existing neighborhoods,” said Ekaterina Savelyeva, senior lecturer in the Department of Architecture and Urban Studies at Perm Polytechnic.
