Abstract
In parallel with the population density in cities, noise, traffic congestion, parking problems and environmental pollution also increase. To address these problems, smart transportation and traffic systems have emerged, which benefit from internet technologies to offer solutions that concern nearly everyone. These systems generate a vast amount of data, often analyzed through machine learning methods. This study has utilized the Adaboost Regression method from the ensemble methods family within the machine learning framework to predict a smart city's traffic model. This method is a combination of many weak learners randomly selected from the data set and created by applying machine learning algorithms to form a strong learner. The Adaboost Regression method has been applied on a smart city traffic models data set found in the Kaggle database. This data set consists of a total of 48,120 rows and 4 columns, including variables such as the number of vehicles, number of intersections, date and time, and ID number. New variables have been created from the date and time variable before starting to analyze the data. The analyses performed with the Adaboost Regression method were carried out in Orange, a free Python-based program. Performance indicators such as Mean Square Error (MSE), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and the coefficient of determination (R2) have been used in the study. A 10-fold cross-validation method was used to ensure the validity of the model and to avoid overfitting. The analysis resulted in an MSE value of 24.19; RMSE value, 4.91; MAE value, 3.00; and R2, 0.94. In conclusion, it has been observed that the AdaBoost Regression method performs successful predictions with low error rates. The Adaboost Regression method, which estimates with minimum error, is also recommended for applications in areas such as smart grid, smart hospital, and smart home, in addition to smart traffic prediction.