| Summary: | 碩士 === 中華大學 === 運輸科技與物流管理學系碩士班 === 99 === To predict traffic volume, data collection usually cost a lot of human power, time and money. In this study, a macroscopic dynamic traffic competition model based on the Lotka-Volterra model is developed so as to predict passenger volume of the public transportation in Taiwan. This study considers four kinds of transportation services, which are high-speed rail, airline, rail road and intercity coach. Firstly, different dynamic traffic prediction models are compared and evaluated by the R-square value. The data is classified into three kinds of time interval. In addition, the oil price is considered as an external variable to describe the social-economical variation. According to results of the analysis, the Mean Absolute Percentage Error (MAPE) and the R-squared value are employed to choose reasonable models. The thresholds of both criteria are 50% and 0.8, respectively. According to the criteria, six possible models are discussed. By analyzing the coefficients of the interaction term, four models among the six possible models are reasonable. However, the results are quite different with the employing data and the external variables. Therefore, the social-economic variables and the studying time period should be analyzed in detail in further researches.
|
|---|
