A major problem with Indian traffic signals is the fact they are hard coded with the timeouts. i.e. The signal has the same timeout throughout the day regardless of the traffic condition. I have proposed two different approaches to tackle the same issue with machine learning. In the first case we predict the timeout of the signal based on previous training data, in the second we classify the amount of traffic at the time using previous training data. This would help improve traffic flow and hence increase efficiency of our roads.
Published In : IJCAT Journal Volume 3, Issue 7
Date of Publication : July 2016
Pages : 378-380
Figures :03
Tables : 02
Publication Link :Using Machine Learning to Improve Signal Timeouts
Sushant Mathur : Department of Computer Engineering, PICT, Savitribai Phule Pune University, Company
Pune, Maharashtra 411043 , India
Thus I have demonstrated how we can use machine learning to learn the patterns in traffic and predict the best possible timeout for a signal. This helps in improving traffic flow and also predicting the amount of traffic at a signal at a given time of day and type of area.
[1] http://www.academia.edu/23089988/The_Econo
nomic_Cost_of_Traffic_Congestion_in_India
[2] L. Kozma, k nearest neighbor algorithm, Helsinki University of Technology. Available : www.lkozma.net/knn2.pdf, 2008
[3] Antti Ajanki AnAj - Own work, CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php? curid=2170282
[4] Rokach, Lior; Maimon, O. (2008). Data mining with decision trees: theory and applications. World Scientific Pub Co Inc. ISBN 978-9812771711.
[5] Altman, N.S.(1992) “An introduction to kernel and nearest neighbor nonparametric regression”. The American Statistician 46(3):175-185. DOI: 10.1080/00031305.1992.1047579
[6] Breiman, Leo; Friedman, J. H.; Olshen, R. A.; Stone, C. J. (1984). Classification and regression trees. Monterey, CA: Wadsworth
