Hazard-based duration modeling (survival analysis) provides addi-tional insights into the underlying duration problem.

Mission

Phone: (716) 645-2114

Fax: (716) 645-3733

Email: panastas@buffalo.edu

The Engineering Statistics and Econometrics Application (E-SEA) research lab was created to support decision making for Engineers and Scientists in problems involving statistical and econometric methodologies and modeling.

The E-SEA lab is in line with the goals of the Institute for Sustainable Transportation and Logistics (ISTL) and the Transportation Informatics University Transportation Center (TransInfo) in regard to data analysis, and its mission is to provide expertise in the field of statistical and econometric methods, modeling, and applications for engineering and applied sciences problems.

To that end, the E-SEA lab provides support to Engineers and Scientists of the School of Engineering and Applied Sciences at UB, and to Governmental Agencies in the State of New York, by offering free consultation on engineering and applied sciences problems that deal with any sort of advanced statistical and econometric modeling analysis.

For a limited time, the E-SEA lab will also provide free consultation to Public and Private Organizations for reasonably sized problems involving statistics or econometrics.

Examples of our modeling approaches include the following:

- Continuous data models (linear and non-linear regression, OLS, WLS, tobit).

- Duration data models (exponential, log-logistic, Weibull, Gamma heterogeneity, etc., hazard-based duration models, survival analysis).

- Count data models (Poisson, negative binomial, zero-inflated models, truncated models).

- Discrete outcome data models (binary and multinomial logit and probit models, nested models, mixed logit models).

- Ordered data models (ordered probability models).

- System of equations models (SURE, 2SLS, 3SLS, multivariate models).

- Models and methods (random parameters modeling, random effects, fixed effects, etc.) dealing with misspecification issues (selectivity bias, unobserved heterogeneity, omitted variable bias, irrelevant variable bias, spurious correlation, heteroskedasticity, multicollinearity, autocorrelation, endogeneity, temporal and spatial parameter transferability, etc.).

Some of the software we use: NLOGIT (LIMDEP), SAS, R, Stata, GeoDA, Mathematica, Mplus, Minitab, WinBUGS, etc.

To contact us:

Discrete outcome modeling is one of the specialty areas of the E-SEA lab.

University at Buffalo, The State University of New YorkReaching Others