No open-access location is known for this record yet.
No open-access location is known for this record yet.
Trip generation in four step travel forecasting model uses linear regression model However, linear regression model has several invalid characteristics. First, trip generation frequencies generally show discrete distribution but the model is estimated based on the assumption that dependent variable has the normal distribution. Second, the model may be able to provide negative values of dependent variable. Third, even if the forecasted value is within valid ranges, only trip frequencies as estimated result can be offered but not discrete probability distribution. To overcome these shortcomings, count data models such as poisson regression or negative binomial regression can be applied. Poisson regression model with a assumption of equidispersion has frequently been used to analyze count data. However, if there is overdispersion in data. poisson regression model tends to underestimates errors, resulting in problem in reliability. In the study applying this concept into analysis on non-home based trip generation in household, after overdispersion testing, it tries to find the possibility to use negative binomial regression model considering overdispersion and improving the reliability. Also, through likelihood test and Theil Inequality Coefficient test, the model is checked for its reliability and accuracy.
Loading…
Imported from OpenAlex (CC0) — nothing here is generated.