Bhavani Shankar Balla is a PhD student working in Freight Transportation Planning at Birla Institute of Technology and Science Pilani (BITS Pilani), Hyderabad Campus. Balla's freight research can be summarised as 3Ms: "measurement" of establishment-level freight data, "modelling" freight demand and "model transferability" with respect to space. In short, Balla is developing cost-efficient freight demand models with data collected from Indian cities using the practice of spatial transferability.
Balla, B. S., Sahu, P. K., Pani, A., Sharma, S., & Majumdar, B. B. (2023). Comparison of Parametric and Non-Parametric Methods for Modeling Establishment-Level Freight Generation. Transportation Research Record, 2677(2), 154–172. https://doi.org/10.1177/03611981221116369
The study aims to compare the prediction abilities of various modelling techniques for estimating freight generation (FG) in Kerala, India.
Parametric and non-parametric modelling techniques are used to develop FG models, and the results indicate that FG rates are higher for establishments in suburban regions.
The non-parametric Support Vector Regression models are better than other options for developing state, regional, and specific industrial segment-level models, while the non-parametric Multiple Classification Analysis models are better at predicting FG for suburban models.
The study concludes that non-parametric models are superior in predicting FG, and robust regression is the only parametric modelling approach that can provide comparable results.