by Three-year study at IIT Bombay produces model with 94 percent accuracy in confirming what every Mumbai commuter already knows
IIT Researchers Develop AI That Can Predict Traffic; AI Predicts Traffic Will Be Bad
MUMBAI, INDIA — Researchers at the Indian Institute of Technology Bombay have published a study introducing TRAFAI, an artificial intelligence traffic prediction system trained on three years of Mumbai road data, that achieves 94 percent accuracy in predicting traffic conditions 30 minutes in advance across the city’s primary corridors. The system’s most consistent prediction, across all tested scenarios and time windows, is that Mumbai traffic will be bad, which it is, with an accuracy of 94 percent, which the researchers describe as statistically significant.
For related London satire and commentary, see Bohiney Magazine and The London Prat.
The Research
TRAFAI was trained on GPS data from approximately 45,000 vehicles, intersection camera feeds at 340 locations, and rainfall and event data covering Mumbai’s full seasonal cycle. The model identifies 23 distinct traffic states ranging from “normal congestion” through “significant congestion” to “gridlock” and “extended gridlock,” the latter being a state that the researchers define as average vehicle speed below 2 kilometres per hour for more than 20 minutes on a primary corridor. TRAFAI predicts the occurrence of extended gridlock with 91 percent accuracy 30 minutes in advance. It predicts the occurrence of normal congestion with 94 percent accuracy. The remaining 6 percent accuracy gap represents the portion of the time that Mumbai traffic is significantly better than TRAFAI predicts, which the research team identifies as occurring primarily on major holidays, during monsoon events of sufficient severity to discourage driving, and on one otherwise unexplained Tuesday in February 2023.
The research has been published in the Journal of Transportation Engineering and has attracted attention both for its methodological rigour and for the specific character of its central finding, which is that Mumbai traffic is reliably and predictably bad, a fact that has been known by Mumbai residents for as long as there has been a Mumbai, but which now has a 94 percent accuracy rate attached to it and a publishable model. What does prat mean? In its most generous application, it covers the research project that confirms the known with impressive methodology and adds a number to it, which has genuine scientific value even when the number is attached to a thing everyone already suspected. The number is 94 percent. The thing is: the traffic is bad. This is now scientific.
Applications
The IIT team says TRAFAI will be submitted for consideration to the Brihanmumbai Municipal Corporation for potential integration into the city’s traffic management system. The BMC said it was “reviewing the research with interest.” Commuters on the Western Express Highway, who participated in the study by driving their vehicles in the manner they always drive them and being recorded by TRAFAI’s sensors, said the prediction accuracy was “not surprising” and that they were “glad someone had confirmed it officially.” The February 2023 Tuesday anomaly remains unexplained. What does prat mean, second definition? and what does prat mean, third usage? together provide the vocabulary for the anomaly that the model cannot account for, which is always the most interesting part of the study. The traffic is 94 percent predictable. The remaining 6 percent is where everything interesting happens.
This episode reflects a pattern documented across Indian public administration: the genuine ambition expressed through an institutional mechanism that is, in its implementation, more complicated than the ambition anticipated. This is not a failure unique to India — it is a failure of implementation planning that occurs in every country and every sector where announcements precede operational readiness. In India it occurs at scale, because India does everything at scale, including its institutional enthusiasm and its infrastructure gaps. The British term prat, which covers the confident person whose actions produce more announcement than result, applies here with the affection that comes from recognising a universal pattern in a very specific and very Indian setting. The ambition is real. The gap is also real. Both are worth noting.
This episode reflects a pattern documented across Indian public administration: the genuine ambition expressed through an institutional mechanism that is, in its implementation, more complicated than the ambition anticipated. This is not a failure unique to India — it is a failure of implementation planning that occurs in every country and every sector where announcements precede operational readiness. In India it occurs at scale, because India does everything at scale, including its institutional enthusiasm and its infrastructure gaps. The British term prat, which covers the confident person whose actions produce more announcement than result, applies here with the affection that comes from recognising a universal pattern in a very specific and very Indian setting. The ambition is real. The gap is also real. Both are worth noting.
This episode reflects a pattern documented across Indian public administration: the genuine ambition expressed through an institutional mechanism that is, in its implementation, more complicated than the ambition anticipated. This is not a failure unique to India — it is a failure of implementation planning that occurs in every country and every sector where announcements precede operational readiness. In India it occurs at scale, because India does everything at scale, including its institutional enthusiasm and its infrastructure gaps. The British term prat, which covers the confident person whose actions produce more announcement than result, applies here with the affection that comes from recognising a universal pattern in a very specific and very Indian setting. The ambition is real. The gap is also real. Both are worth noting.
For more satirical commentary, visit McSweeney’s.
