| Summary: | This project is aimed to study the relationship between the number of available taxi, time-of-theday, road location and speed band, as well as the intensity prediction of taxi availability given different variables using the implementation of Naïve Bayes Classifier. LTA have tried to increase the number of taxi plying on roads in hopes for passengers to flag a taxi with ease. However, there are still feedbacks and complaints from users the difficulty in hiring a taxi within reasonable waiting time, especially during peak hours. The two main reasons are due to the introduction of app-based taxi (Uber, Grab) and the imbalance distribution of taxi supply in Singapore. In this report, the focus will be on the latter. Real-time data collection of taxi availability as well as the speed band of different roads are gathered using LTA API call and python programming. Further correlation between the two dataset is programmed to visualise the number of available taxi on different roads indicating its time-of-the-day and speed band. Using the enhanced data, variables are categorised and run in a Naïve Bayes Classifier program to predict the intensity of taxi given different variables. The results of the program will enable us to analyse the imbalance distribution of taxi supply. In conclusion, areas with high human traffic/activity will experience high taxi intensity and lower waiting time while areas with low human traffic/activity will experience otherwise. Areas with low taxi intensity will then resort to other alternatives such as app-based taxi or other modes of public transport as the waiting time to hire a taxi can be too long. It is also important to take note that there is a dynamic supply and demand of taxi meaning that the numbers may change depending on time, traffic condition, location etc. Recommendations have been made to further analyse the ratio of supply and demand within each road to have a supply-demand balance in each area rather than Singapore as a whole.
|
|---|