|
Macro or micro | Evaluation methods | Specific methods | Advantages | Disadvantages |
|
Macro | Method based on bike-sharing and public transit ridership data | (i) DID model | Ability to analyze the relationship between public transit ridership and bike-sharing ridership (or facility supply), as well as the relationship between the intensity of public transit facility supply and bike-sharing ridership, at the macrolevel | Inability to analyze the substitution, connection, and complementation relationships between bike-sharing and public transit at the microlevel of individual trip |
(ii) Linear regression model considering error autocorrelation |
(iii) OLS model |
(iv) Synthetic control method |
(v) Random forest model |
(vi) Interrupted time series model |
(vii) Bayesian structural time series model |
(viii) Complex network theory approach |
|
Micro | Method based on bike-sharing user behavior surveys | (i) Statistical analysis of the proportion of connection and substitution trips | Ability to analyze the actual connection and substitution relationships between bike-sharing and public transit at the microlevel of individual trips | Small sample size, high survey costs, and inability to cover the trips of all individuals |
(ii) Statistical analysis of the change in frequency of users’ use of original travel modes before and after the emergence of bike-sharing |
(i) Binomial logit model | Ability to identify the factors influencing the coopetition relationship between bike-sharing and public transit | The structure and assumptions of these models are different and their applicability should be examined in different research scenarios |
(ii) Multinomial logit model |
(iii) Hybrid logit model |
|
Micro | Method based on bike-sharing transaction data | (i) Identification methods of potential coopetition relationships based on spatial-temporal relationship between bike-sharing rides and public transit | Ability to quickly obtain large or even complete samples of transaction data at low cost for analysis of coopetition relationships | Only potential coopetition relationships can be identified, and the accuracy of the identification has not yet been evaluated |
|