Passengers Travel Behavior Research Based On The Data Mining Technology

With the accelerating pace of urbanization construction, China’s urbanization rate in 1978 from 17.9 percent to 56.1 percent in 2015, the urban resident population exceeded 770 million. People travel demand is also growing, showing rapid growth in road traffic, especially in Beijing, Shanghai and other big cities, such as traffic congestion, travel inconvenience and other problems are constantly increasing, these problems have seriously affected the city dwellers the normal work and life, restricting the healthy development of the city. Public transport as an urban transport important part in the city has been a great development, but in most cities the number of public transport trips than the total travel is still low, in part due to public transport planning law and policy development and the relationship between the traveling public, there are still inconsistent.Data mining is an important research of intelligent systems that can tap hidden in one of the law and knowledge from large amounts of data. Based on historical bus card transactions some bus lines in Guangdong Province, through data mining technologies and processes for fixing crowd behavior on public transport in the mining, analysis speculate passenger travel habits and preferences, the final model to predict people’s future week fixed bus routes travel situation, which is of great significance for the majority of passengers with information symmetry and reasonable and safe travel environment.Firstly, data mining theoretical background and method of doing a general presentation and analysis, based on all the data, from the perspective of practical business point of view, on the passenger presence influence behavior regularity patterns in the fixed bus lines future travel behavior made some inquiry proposed construction method of a reaction passengers historical behavior and preferences of the engineering system features. According to the characteristics of the feature engineering design, as well as to the random forest and gradient iteration decision tree theory analysis and application of the combination of two kinds of learning algorithm, this paper established a better robustness for citizens to travel to forecast model in the future, and based on this, we have two kinds of combination algorithm made a series of analysis and comparative study...