Abstract
As mobile ride-hailing applications become increasingly popular, the "Internet +" model has transformed the taxi industry. While these platforms offer convenience through subsidies, they also face challenges in efficient resource allocation. This study develops mathematical models to address two primary issues: matching taxi supply and demand and optimizing subsidy policies. Using a fuzzy comprehensive evaluation model, we identify the "load factor" as a critical indicator influenced by time and space. Furthermore, we establish a multiple regression model using SPSS to determine the relationship between subsidies, taxi ownership, and user populations, providing a framework for designing sustainable subsidy schemes.
Introduction
Despite the convenience of mobile ride-hailing apps, the "difficulty of taking a taxi" remains a persistent social challenge. In the "Internet +" era, taxi companies use massive subsidies to attract drivers and passengers, but these are often applied haphazardly, leading to inefficient resource allocation. This research tackles two main questions: how to mathematically define and evaluate the match between taxi supply and demand in specific regions, and how platforms can design sustainable subsidy policies that avoid market imbalances.
Methodology
The researchers employed a two-pronged mathematical approach to analyze the taxi market dynamics.
Fuzzy Comprehensive Evaluation
To measure the "demand match degree," three key indicators were established: mileage utilization rate, load factor, and taxi ownership per 10,000 people. A judging matrix was used to categorize results into "range," "medium," or "favorable" states.
Spatiotemporal Modeling
Recognizing that demand is dynamic, the study modeled the time factor using a standard normal distribution probability density function to identify peak windows (e.g., 7:00–8:00 AM). The space factor was modeled using scatter plot analysis to visualize demand clusters.
The analysis determined that taxi capacity should be inversely proportional to the distance ($d$) from a central point ($1/d$), following a "center divergence" principle where demand radiates outward from specific hubs.
Multiple Linear Regression
To solve the subsidy optimization problem, a regression model was created using SPSS. This model treats the total cost of subsidies as a dependent variable influenced by the proportion of the population taking taxis and total taxi ownership.
Data Analysis
The study primarily utilized historical data from Beijing between 2012 and 2015:
- Beijing Taxi Mileage Data (2012–2015): Average monthly mileage and passenger miles used to calculate utilization rates.
- Demographic and Infrastructure Data: Beijing’s population (19.72 million) and downtown taxi ownership (66,646).
- Subsidy Records: A dataset tracking total subsidy costs, taxi ownership, and market penetration from January 2013 to September 2015.
Results and Discussion
The fuzzy evaluation revealed that the load factor (the ratio of carrying time to total time) is the most volatile indicator, heavily influenced by both time and space. This makes it the most critical metric for real-time resource adjustment.
The mileage utilization rate in Beijing remained stable at approximately 68% from 2012 to 2014 but dropped sharply to 31.27% in early 2015, indicating a significant market disruption.
The study produced the following predictive model for subsidies: $$\hat{y} = -36391.8 + 0.053a_1 + 0.885a_2$$ Where $a_1$ represents the population proportion and $a_2$ represents taxi ownership. With an F-statistic of 19.370, the model demonstrates practical statistical significance for estimating necessary subsidy budgets based on target market penetration.
Conclusion
While the regression model provides a useful framework for subsidy planning, limitations remain. Data processing bottlenecks restricted the number of data points used, potentially limiting the model's universality. Future work should incorporate environmental factors (weather, seasons) and passenger demographics (income, travel purpose) to improve demand forecasting. Additionally, this regression approach could be extended to other commodity subsidy contexts beyond the taxi industry.
Summary updated